ABSTRACT

Title of Dissertation: Novel Organic Polymeric and Molecular

Thin-Film Devices for Photonic Applications

Younggu Kim, Doctor of Philosophy, 2006

Dissertation directed by: Professor Chi H. Lee

Department of Electrical and

Computer Engineering

The primary objective of this thesis is to explore the functionalities of new classes

of novel organic materials and investigate their technological feasibilities for be-

coming novel photonic components.

First, we discuss the unique polarization properties of optical chiral waveguides.

Through a detailed experimental polarization analysis on planar waveguides, we

show that eigenmodes in planar chiral-core waveguides are indeed elliptically po-

larized and demonstrate waveguides having modes with polarization eccentricity

of 0.25, which agrees very well with recent theory. This is, to the best of our

knowledge, the first experimental demonstration of the mode ellipticities of the

chiral-core optical waveguides. In addition, we also examine organic magneto-optic

materials. Verdet constants are measured using balanced homodyne detection, and

we demonstrate organic materials with Verdet constants of 10.4 and 4.2 rad/T · m

at 1300 nm and 1550 nm, respectively.

Second, we present low-loss waveguides and microring resonators fabricated

from perfluorocyclobutyl copolymer. Design, fabrication and characterization of

these devices are addressed. We demonstrate straight waveguides with propagation

losses of 0.3 dB/cm and 1.1 dB/cm for a buried channel and pedestal structures,

respectively, and a microring resonator with a maximum extinction ratio of 4.87 dB,

quality factor Q = 8554, and finesse F = 55. In addition, from a microring-loaded

Mach-Zehnder interferometer, we demonstrate a modulation response width of

30 ps and a maximum modulation depth of 3.8 dB from an optical pump with a

pulse duration of 100 fs and a pulse energy of 500 pJ when the signal wavelength

is initially tuned close to one of the ring resonances.

Finally, we investigate a highly efficient organic bulk heterojunction photode-

tector fabricated from a blend of P3HT and C60. The effect of multilayer thin

film interference on the external quantum efficiency is discussed based on numer-

ical modeling. We experimentally demonstrate an external quantum efficiency

ηEQE = 87 ± 2% under an applied bias voltage V = −10 V, leading to an internal

quantum efficiency ηIQE ≈ 97%. These results show that the charge collection

efficiency across the intervening energy barriers can indeed reach near 100% under

a strong electric field.

Novel Organic Polymeric and Molecular Thin-Film Devices

for Photonic Applications

by

Younggu Kim

Dissertation submitted to the Faculty of the Graduate School of the

University of Maryland, College Park in partial fulfillment

of the requirements for the degree of

Doctor of Philosophy

2006

Advisory Committee:

Professor Chi H. Lee, Advisor/Chair

Doctor Warren N. Herman, Co-advisor

Professor Julius Goldhar

Professor Christopher C. Davis

Professor James R. Anderson

c© Copyright by

Younggu Kim

2006

DEDICATION

To the loving memory of my father,

and my mother

ii

ACKNOWLEDGMENTS

I am truly grateful for the advice, support, friendship, criticism, and collaboration

that I have received over the years during my graduate study at the University of

Maryland and my research on polymer photonics at the Laboratory for Physical

Sciences. LPS has been a really fascinating place to conduct research because of

the wonderful facilities and equipment as well as the many talented scientists and

engineers.

First of all, I am particularly indebted to Prof. Chi Lee for his advice through-

out my Ph.D research, and the tremendous amount of support and encouragement

when I would get stuck. His great enthusiasm for research and for playing ten-

nis impressed me often considering his age, and he sometimes reminded me of

my father who would be the same age as he. I am deeply grateful to Dr. Warren

Herman at the LPS who introduced me to this wonderful area of research on poly-

mer photonics and optoelectronics. He has been extremely patient in answering

my countless questions. Without a doubt, every single piece of knowledge on or-

ganic polymer optics I have acquired has benefited from daily interaction with him.

Along the way of my Ph.D study, I was lucky enough to have a second academic

advisor, Prof. Julius Goldhar. I am grateful for his fruitful insights and comments

on the experiments I have conducted. I will miss his sense of humor as well. I

iii

am also grateful to Prof. Chris Davis and Prof. Bob Anderson who agreed to serve

on my dissertation committee. When I was new to optics, I learned a lot through

classes I took from Prof. Davis. He was also generous enough to extend my final

exam when I had to go back to Korea for my father’s funeral.

I also would like to thank Prof. Victor Granastein who gave me a research

opportunity for the first year of my graduate study. He is a genuine gentleman

whom everybody loves. I deeply appreciate his support even after I left his group.

I have also benefited from many stimulating discussions with Prof. Tom Murphy on

a few numerical simulations. His Matlab codes were exceptionally helpful to design

ring resonator devices. In addition, his class on “Optical Communication Systems”

was definitely the best among the classes I took at the University of Maryland. I

also want to thank Dr. Danilo Romero, an expert on organic optoelectronic devices.

He not only fabricated a series of devices for my research but also patiently taught

me a lot on the physics of organic semiconductors.

All the thesis work described here would not have been possible without the

support and encouragement of my colleagues at the University of Maryland and

LPS: Toby Olver, Dan Hinkel, Steve Brown, Scott Horst, Lisa Lucas, Sal Mar-

tinez, Mike Khbeis, George de la Vergne, J.B. Dottellis, Russell Frizzell, Leslie

Lorenz, Mark Thornton, Greg Latini, Dr. Chris Richardson, Dr. Charles Krafft,

Prof. P. T. Ho, Dr. Rohit Grover, Dr. Vien Van, Dr. Kuldeep Amarnath, Victor

Yun, Dr. Yongzhang Leng, Dr. Paul Petruzzi, Dr. Wei-Lou Cao, Dr. Jungwhan Kim,

Min Du, and Yi-Hsing Peng, and more. Special thanks go to cleanroom manager

iv

Toby who has never turned down my request for help and made my work in the

cleanroom a lot easier, Dan who was truly helpful for device fabrication and SEM-

ing, Steve whom I bothered almost every day for stepper use, Dr. Cao who helped

me conduct various experiments, and Dr. Krafft who provided useful tips on the

experimental setup to measure the Faraday rotation.

Most of the work in my thesis has been done in strong collaboration with other

researchers outside of the University of Maryland and LPS. I would like to thank

Prof. Mark Green, Dr. Myungjin Lee and Dr. Henry Teng at N.Y. Polytechnic Uni-

versity, Prof. Andrzej Rajca at the University of Nebraska, Prof. Lin Pu at the

University of Virginia, Prof. Dennis Smith and Shengrong Chen at Clemson Uni-

versity, Prof. Seth Marder at Georgia Tech, Dr. Cheng Zhang at Norfolk University,

Dr. William Diffey at the U S Army, Dr. Andy Guenthner at NAWC China Lake.

I am grateful to Shuo-Yen Tseng, Glenn Hutchinson, Mike Bowen, Ricardo

Pizarro, Dyan Ali, Arthur Liu, and more, especially to Shuo-Yen and Glenn for

the time we spent together not only at LPS but also at Hard Times and Starbucks.

I will particularly miss the happy hour a lot. I also would like to thank several

fellow Korean friends in the department including Dong Hun Park, Soo Bum Lee,

Woochul Jeon, Jusub Kim, and Seokjin Kim for the occasional wonderful parties

and the time we spent on the tennis court, and Kyuyong Lee, Jeong-Nam Kim,

Kiyong Kim, Seungjoon Lee, Joonhyuk Yoo, Seung-Jong Baek, Kyechong Kim,

Sangchul Song, Inseok Choi, Hojin Kee, Jookyung Lee, and many more for their

friendship. My graduate study would not have been nearly as much fun without

v

them. Additionally, I want to thank Jeong-Il Oh who did not mind driving down

here from Boston when I was frustrated from work or life.

Most of all, I am deeply indebted to my family for their love, encouragement,

and support (including monetary support, as ‘indebted’ means in a dictionary)

over the years. No matter how many words I put here, it will never be sufficient to

express my true gratefulness to my dad in heaven, mom, parents-in-law, brother,

and sister. Finally, I would like to gratefully acknowledge Soyeon and Minju for

their constant love, sacrifice, and tolerance of my irregular hours at home.

vi

TABLE OF CONTENTS

Acknowledgments iii

List of Figures x

List of Abbreviations xviii

1 Introduction 1

1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Scope of thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Optical Chiral Waveguide 8

2.1 Organic chiral material and optical activity . . . . . . . . . . . . . . 8

2.2 Unique polarization properties of chiral-core planar waveguides . . . 11

2.2.1 Fields in bulk chiral media . . . . . . . . . . . . . . . . . . . 12

2.2.2 Eigenmodes in asymmetric chiral planar waveguides . . . . . 15

2.3 Chiral waveguides from amorphous binaphthyl films . . . . . . . . . 25

2.4 Experimental polarization analysis . . . . . . . . . . . . . . . . . . 34

2.5 Other chiral materials . . . . . . . . . . . . . . . . . . . . . . . . . 48

vii

2.5.1 Binaphthyl polymer . . . . . . . . . . . . . . . . . . . . . . . 49

2.5.2 Polymethine dyes . . . . . . . . . . . . . . . . . . . . . . . . 50

2.5.3 (7)Helicene . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

2.5.4 Bridged binaphthyl with chromophore . . . . . . . . . . . . 52

2.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

3 Organic Magneto-Optic Material 55

3.1 Magneto-optic effect . . . . . . . . . . . . . . . . . . . . . . . . . . 55

3.2 Balanced homodyne detection . . . . . . . . . . . . . . . . . . . . . 58

3.3 Measurement of Verdet constant . . . . . . . . . . . . . . . . . . . . 63

3.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

4 Optical Waveguide and Microring Resonator from PFCB 74

4.1 Perfluorocyclobutyl (PFCB) copolymer . . . . . . . . . . . . . . . . 74

4.2 Low loss optical waveguide from PFCB copolymer . . . . . . . . . . 75

4.2.1 Eigenmodes in dielectric waveguides . . . . . . . . . . . . . 77

4.2.2 Fabrication of PFCB waveguides . . . . . . . . . . . . . . . 80

4.2.3 Loss measurement of straight waveguides . . . . . . . . . . . 84

4.3 PFCB Microring resonators . . . . . . . . . . . . . . . . . . . . . . 87

4.3.1 Basic principles . . . . . . . . . . . . . . . . . . . . . . . . . 88

4.3.2 Design considerations . . . . . . . . . . . . . . . . . . . . . . 94

4.3.3 Experimental characterization . . . . . . . . . . . . . . . . . 100

4.3.4 Ultrafast all-optical switching experiment . . . . . . . . . . . 106

viii

4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

5 Organic Thin-Film Photodetector based on Bulk Heterojunction 117

5.1 Basic principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

5.2 Fabrication of bulk heterojunction PDs . . . . . . . . . . . . . . . . 126

5.3 Device characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . 131

5.4 Multilayer thin film interference effects . . . . . . . . . . . . . . . . 138

5.4.1 Optical properties of individual layers . . . . . . . . . . . . . 138

5.4.2 Effect of multilayer thin film interference . . . . . . . . . . . 140

5.5 Improvement of the external quantum efficiency . . . . . . . . . . . 152

5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

6 Conclusions 158

6.1 Summary and accomplishments . . . . . . . . . . . . . . . . . . . . 158

6.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161

Bibliography 165

ix

LIST OF FIGURES

2.1 Structure of an asymmetric planar waveguide with an isotropic chi-

ral core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.2 Calculation of the mode effective index neff vs. waveguide thickness

d. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.3 Calculation of the mode eccentricity parameter g vs. chirality γ for

the lowest order elliptical modes, RHE0 and LHE0. . . . . . . . . . 23

2.4 Chemical structure of the single molecule binaphthyls, ML-224 and

ML-226. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

2.5 Optical rotatory dispersion (ORD) curves for ML-224 and ML-226

measured in solution. . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.6 Index of refraction dispersion measured with a MetriconTM for ML-

224 and ML-226 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2.7 Experimental setup for measurement of optical rotatory power. . . 29

2.8 Structure of chiral slab waveguides on GaAs substrates coated with

SiON. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

x

2.9 Calculation of neff with increasing waveguide core thickness, (a)

ML-224 and (b) ML-226. . . . . . . . . . . . . . . . . . . . . . . . 32

2.10 Calculation of mode ellipticity of RHE0 and LHE0 of the asymmet-

ric slab waveguide as a function of given material chirality γ. . . . 33

2.11 Experimental setup for polarization analysis for chiral waveguide. . 35

2.12 Detected optical power as a function of rotating quarter waveplate

orientation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

2.13 Continuous evolution of the polarization states on the Poincare

sphere. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

2.14 Evolution of polarization through the chiral waveguide fabricated

from ML-224 and ML-226. . . . . . . . . . . . . . . . . . . . . . . . 44

2.15 Calculation of mode ellipticity of RHE0 and LHE0 as a function of

given material chirality γ for a symmetric and asymmetric chiral-

core planar waveguides. . . . . . . . . . . . . . . . . . . . . . . . . 47

2.16 Chemical structure of the main chain binaphthyl polymer, Zhang-

IV-37 with R = n-C6H13. . . . . . . . . . . . . . . . . . . . . . . . 49

2.17 Index of refraction dispersion measured with a MetriconTM for the

polybinaphthyl Zhang-IV-37 . . . . . . . . . . . . . . . . . . . . . . 50

2.18 UV-Vis-IR spectra measured with a Cary 5000 spectrophotometer

for HT-dye14 in poly(butyl methacrylate-co-methyl methacrylate)

copolymer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

xi

3.1 Schematics of balanced homodyne detection. . . . . . . . . . . . . . 60

3.2 Normalized intensities along the principal axes I1, I2, and difference

I1 − I2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

3.3 Experimental setup of balanced homodyne detection to measure

the Verdet constant. . . . . . . . . . . . . . . . . . . . . . . . . . . 64

3.4 Calibration of the amplitude of alternating magnetic flux density

inside the Helmholtz coil. . . . . . . . . . . . . . . . . . . . . . . . 65

3.5 Optical bias curve without magnetic field applied for a 2 cm-long

rod of TGG sample obtained at wavelength 633 nm . . . . . . . . . 67

3.6 Faraday rotation measured from a 2 cm long rod of TGG at a wave-

length of 633 nm by varying magnetic field flux densities and a fit

to linear line. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

3.7 Faraday rotation measured for a 2 cm-long rod of TGG at different

wavelengths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

3.8 Chemical structure of SJZ-87A . . . . . . . . . . . . . . . . . . . . 69

3.9 UV-Vis-IR transmission spectra measured with a Cary 5000 spec-

trophotometer for organic magneto-optic samples provided by Geor-

gia Tech. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

3.10 Faraday rotation measured for the substrates only and for the sam-

ple labelled SJZ-87A at wavelengths of 1300 nm and 1550 nm. . . . 71

xii

3.11 Faraday rotation measured for the substrates only and for the sam-

ples labelled TK1V1292A320-A and TK1V1292A320-N at a wave-

length of 850 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

4.1 Thermal (a)polymerization and (b)copolymerization to PFCB-based

polymer from trifluorovinylaryl ether monomers . . . . . . . . . . . 76

4.2 Mode profile of the transverse electric field component |ex| for the

fundamental TE mode for a pedestal and buried channel . . . . . . 79

4.3 Outline of process steps to fabricate the PFCB waveguides. . . . . 82

4.4 Scanning electron micrograph of the PFCB pedestal waveguide. . 85

4.5 Loss measurement of PFCB straight waveguides using cutback method

for TE mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

4.6 Schematic diagram of microring resonators. (a) all-pass and (b)

add-drop configuration. . . . . . . . . . . . . . . . . . . . . . . . . 88

4.7 Calculation of a typical spectral response of all-pass filter. . . . . . 91

4.8 Calculated phase response near a resonance in all-pass configuration 92

4.9 Calculation of a typical spectral response of add-drop filter . . . . . 93

4.10 3D full vectorial calculation for bending loss. . . . . . . . . . . . . 96

4.11 Cross-sectional mode profile of two parallel waveguides separated

by an edge-to-edge distance g. . . . . . . . . . . . . . . . . . . . . 98

4.12 Calculation of coupling constant between two PFCB pedestal waveg-

uides as a function of waveguide separation g. . . . . . . . . . . . 99

xiii

4.13 Scanning electron micrograph of PFCB ring resonator, with a ra-

dius R = 25 µm, and a coupling gap g = 0.45 µm. . . . . . . . . . 102

4.14 Experimental setup for microring characterization. . . . . . . . . . 103

4.15 Measured spectral response at the drop port of a add-drop filter

with a radius 25µm with ASE input . . . . . . . . . . . . . . . . . 104

4.16 Normalized signal of the device at the throughput and drop port

at the resonance at 1531.05 nm from a tunable laser input. . . . . 105

4.17 (a) Resonance shift with increasing ASE input power. Maximum

resonance shift 25 pm obtained. (b) Thermo-optic coefficient, dn/dT ,

measured using a MetriconTM that was custom outfitted with tem-

perature control. . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

4.18 Schematic picture of all-optical switching experiment. . . . . . . . . 109

4.19 Spectral response of an MR-MZI with a length imbalance between

two arms of the MZI. . . . . . . . . . . . . . . . . . . . . . . . . . 111

4.20 Measured switching response of the MR-MZI from a optical pump,

a 100 fs Ti:Sapphire laser pulse . . . . . . . . . . . . . . . . . . . . 113

4.21 Calculated phase response near a resonance in all-pass configuration

when the ring is lossy and under-coupled . . . . . . . . . . . . . . . 115

5.1 Schematic drawing illustrating photoinduced charge transfer be-

tween a conjugated polymer (P3HT) and a fullerene (PCBM-C60). 119

xiv

5.2 (a) Schematic energy level diagram of electrodes, donor and accep-

tor material (b) Schematic illustration of photocurrent generation

in heterojunction PV cells and detectors. . . . . . . . . . . . . . . . 122

5.3 Schematic diagram depicting conceptually the photoinduced charge

transfer in (a)bilayer heterojunction and (b)bulk heterojunction . . 124

5.4 The chemical structure of the materials in bulk heterojunction layer

for organic photodetector . . . . . . . . . . . . . . . . . . . . . . . 127

5.5 Schematic diagram of the device configuration. . . . . . . . . . . . 128

5.6 Atomic force microscopy (AFM) images for four different PCBM-

C60 molar fractions x. . . . . . . . . . . . . . . . . . . . . . . . . . 130

5.7 Current density versus voltage (I-V ) characteristics of the bulk het-

erojunction device of MEH-PPV copolymer and PCBM-C60 with a

C60 molar fraction x = 0.57. . . . . . . . . . . . . . . . . . . . . . . 133

5.8 Absorption spectra of the blend of P3HT/PCBM-C60 . . . . . . . . 135

5.9 I-V characteristics of the bulk heterojunction device from P3HT

and PCBM-C60 with a C60/P3HT 1.1 in weight, (a) under AM 1.5

direct solar illumination (b) under different laser intensities at 532 nm.136

5.10 External quantum efficiency vs. bias voltage of the device of P3HT/PCBM-

C60 at wavelength of 532 nm . . . . . . . . . . . . . . . . . . . . . . 137

xv

5.11 (a) Real and imaginary index of refraction measured from the spec-

troscopic ellipsometric data and model fit with the Drude-Lorentz

model. (b) Comparison between the estimated transmission using

the n and k, and thickness from the best-fit model and the measured

transmission from a spectrophotometer. . . . . . . . . . . . . . . . 141

5.12 Optical properties of Al, PEDOT:PSS, and P3HT:C60 . . . . . . . 142

5.13 Schematic diagram of the multilayer device structure with the thick-

nesses and the optical electric fields in each layer, E+i and E−

i . . . . 143

5.14 Calculation of reflectivity RD and absorption AD = 1 − RD − TD

of the bulk heterojunction device of P3HT/C60 with a thickness of

100 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

5.15 Calculation of field distribution |E(x)|2 as a function of position.

Modulus squared of the optical electric field |E(x)|2 is normalized

to the incident light field |E+0 |2. . . . . . . . . . . . . . . . . . . . . 148

5.16 Calculation of time-averaged absorbed power Q(x) as a function of

position. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

5.17 Calculation of field distribution |E(x)|2 and time-averaged absorbed

power Q(x) as a function of position for the device structure of

glass/ITO/P3HT:C60/Al. . . . . . . . . . . . . . . . . . . . . . . . 153

5.18 UV-Vis-IR transmission spectra for P3HT:C60 samples with three

different thicknesses. . . . . . . . . . . . . . . . . . . . . . . . . . . 154

xvi

5.19 Calculated upper limit of the external quantum efficiency ηLEQE as

a function of active layer thickness dA and the measured EQE at

bias V =-10 V, with and without PEDOT:PSS and LiF layers. . . . 155

5.20 Measured ηEQE as a function of applied reverse bias voltage (a) and

photocurrent at V = −10 V as a function of incident laser intensity

(b) for a device with a dA = 300 nm. . . . . . . . . . . . . . . . . . 157

xvii

LIST OF ABBREVIATIONS

TE Transverse electric

TM Transverse magnetic

RHC Right-handed circular

LHC Left-handed circular

RHE Right-handed elliptical

LHE Left-handed elliptical

SOP State of polarization

PECVD Plasma enhanced chemical vapor deposition

RIE Reactive ion etching

ICP-RIE Inductive coupled plasma RIE

SEM Scanning electron microscope

TGG Terbium gallium garnet

PFCB Perfluorocyclobutyl

BCB Bisbenzocyclobutene

CTE Coefficient of thermal expansion

FD Finite difference

PDL Polarization dependent loss

xviii

PML Perfectly matched layer

OCDF Optical channel dropping filter

OADF Optical add-drop filter

MZI Mach-Zehnder interferometer

MR-MZI Microring-loaded Mach-Zehnder interferometer

ASE Amplified spontaneous emission

EDFA Erbium doped fiber amplifier

FWHM Full width at half maximum

OPV Organic photovoltaic

OPD Organic photodetector

HOMO Highest occupied molecular orbital

LUMO Lowest unoccupied molecular orbital

OLED Organic light emitting diode

EQE External quantum efficiency

ICPE Incident photon to current conversion efficiency

IQE Internal quantum efficiency

PCE Power conversion efficiency

ITO Indium tin oxide

P3HT Poly(3-hexylthiophene-2,5-diyl)

PCBM-C60 [6,6]-phenyl-C61-butyric acid methyl ester

PEDOT Poly(3,4-ethylenedioxythiophene)

PSS Poly(styrenesulfonate)

xix

Chapter 1

Introduction

1.1 Motivation

Organic polymers and molecules have emerged as one of the most promising class of

materials for photonic and electronic applications due to their potential capabilities

and advantages such as moderate to high functionality, fabrication flexibility, sub-

strate compatibility, affordability, etc. While other material systems, such as inor-

ganic semiconductors (e.g., silicon, gallium arsenide, and indium phosphide, etc.),

glasses (e.g., silicon dioxide, silicon oxynitride, and silicon nitride, etc.), silicon

on insulator (SOI), lithium niobate, sol-gels, etc. have widely been used, organic

materials have become a new paradigm for various components in communication

systems, displays, sensors, solar cells, and so on. During the past decade, there has

been an explosive growth and development of both materials and devices. Repre-

sentative examples include waveguide based integrated-optic devices for essential

building blocks for optical communication systems based on linear/nonlinear op-

tical properties of the materials: interconnectors, polarization converters, filters,

switches, directional couplers, modulators, gratings [1, 2]. They also include opto-

electronic and electronic devices based on electroluminescent and (semi)conducting

1

properties of the materials: LEDs, photodetectors, photovoltaic cells, lasers, radio-

frequency identification (RFID) tags, field effect transistors (FETs) [3, 4].

Organic materials have great advantages in many aspects because of the abil-

ity to change the material properties synthetically by modifications in chemical

structure and composition. Advantages include:

• tunability of index of refraction (range of 1.3 – 2) and birefringence control

• low coupling loss from and to the fiber due to small Fresnel reflection and

mode mismatch

• low absorption loss at common telecommunication wavelength, 1310 nm and

1550 nm

• large class of 2nd and 3rd order nonlinear optical properties

• various optoelectronic properties such as photoluminescence, electrolumines-

cence, and (semi)conductivity

• easily formed into thin films via spin-coating, inkjet printing, dipping, etc.

• wide variety of substrate compatibility, such as glass, quartz, semiconductor,

printed circuit board, and even flexible ones

• fabrication flexibility – photolithography as well as non-conventional lithog-

raphy such as laser ablation, nano-imprinting, multi-photon absorption, etc.

2

• affordability – inexpensive materials, simple fabrication processes, and fast

turn-around time

On the other hand, organic devices have drawbacks in terms of environmental

performance and reliability, such as thermal aging and photo-oxidation. Many

research groups, including electrical engineers, chemists, and physicists, have been

making efforts to overcome such problems through further chemical modification

and extensive reliability testing, such that devices in some applications are already

in the market (OLED, for example) and some are soon to be commercially available.

New classes of organic materials for various applications are synthesized ev-

ery day. The primary objective of this thesis is to explore the functionalities

of those materials and investigate their technological feasibilities for constructing

novel photonic components.

1.2 Scope of thesis

This thesis is separated into three principal parts discussing different types of

devices that are somewhat independent of each other, but can be integrated into

single devices: chiral waveguides and magneto-optic materials, low-loss waveguides

and microring devices from perfluorinated polymer, and organic bulk heterojunc-

tion photodetectors.

Chapter 2 will discuss the unique polarization properties of optical chiral waveg-

uides that can be used for integrated-optic polarization converters or polarization

modulators provided that low loss amorphous films of sufficient chirality can be

3

fabricated. In an achiral film, as is well known, the modes are transverse electric

(TE) and transverse magnetic (TM). On the other hand, new modeling results for

(a)symmetric planar waveguides with chiral cores showed that the eigenmodes are

elliptically polarized modes, in general, with the polarization ellipticity depend-

ing on the chirality, which is related to the bulk rotatory power of material. For

an important design consideration, dependence of mode ellipticity on cladding in-

dex is also discussed, and we propose and demonstrate a SiON layer for a cladding

layer. Through a detailed experimental polarization analysis on asymmetric chiral-

core planar waveguides fabricated from binaphthyl-based organic single molecules,

the eigenmodes of chiral planar waveguides are characterized and their mode el-

lipticities are compared with the ones predicted in recent theory. We show that

eigenmodes in planar chiral-core waveguides are indeed two orthonormal elliptical

polarizations and demonstrate waveguides having modes with a polarization eccen-

tricity of 0.25, which agrees very well with the theory. Although the achieved mode

ellipticity is not large enough to meet the requirement – predominant circularly

polarized modes – for practical applications, this is, to the best of our knowledge,

the first experimental demonstration of the mode ellipticities for the transverse

electric field of the chiral-core optical waveguides. Characterizations and feasibil-

ities of several other chiral materials are also discussed. Unfortunately, none of

these materials exhibit both appreciable material chirality and negligible linear

birefringence at the same time. The design and synthesis of isotropic materials

with high chirality is in progress.

4

Chapter 3 will examine organic magneto-optic materials, which can be viable

alternatives to the chiral materials. In addition to polarization rotators, they also

can potentially be used for integrated-optic isolators or circulators. The Faraday

effect in magneto-optic materials in a magnetic field is another class of optical activ-

ity, and it is characterized by Verdet constant. Measuring the Verdet constants of

thin (10−100 µm) organic samples under a moderate magnetic field (< 100 Gauss)

can be very challenging. For this reason, we discuss balanced homodyne detection

that provides a highly sensitive technique to measure an angle of polarization ro-

tation as small as 0.5 × 10−6 rad. We demonstrate the Verdet constants of 10.4

and 4.2 rad/T · m at 1300 nm and 1550 nm, respectively, from an organic sample

provided by Georgia Tech, which is comparable to that of terbium gallium garnet.

This unique observation can lead to integrated-optic isolators or circulators from

a simple fabrication methodology including spin-casting and photolithography.

Chapter 4 will present low-loss waveguides and polymer microring resonators

fabricated from perfluorocyclobutyl (PFCB) copolymer. Among the many materi-

als for passive waveguide devices, fluorinated polymers are particularly attractive

because they have very low absorption loss at telecommunication wavelengths.

Low-loss waveguides are one of the most important building blocks in integrated-

optic communication systems because most passive and active devices consist of

simple straight and curved waveguide segments. For example, microring resonator

based devices have very simple configurations of straight waveguides and circular

rings, and the loss present in the ring is the most important parameter deter-

5

mining the device performance. For these reasons, this chapter is devoted to the

design, fabrication, and characterization of those devices using PFCB. We demon-

strate straight waveguides with propagation losses of 0.3 dB/cm and 1.1 dB/cm for

a buried channel and pedestal structures, respectively, and a microring resonator

with a maximum extinction ratio of 4.87 dB, quality factor Q = 8554, and finesse

F = 55. In addition, we demonstrate that all-optical switching with the PFCB

microring resonator is possible when it is optically pumped with a femtosecond

laser pulse of a sufficient energy. From a microring-loaded Mach-Zehnder inter-

ferometer, we demonstrate a modulation response width of 30 ps and a maximum

modulation depth of 3.8 dB from a optical pump with a pulse duration of 100 fs

and pulse energy of 500 pJ when the signal wavelength is initially tuned close to

one of the ring resonances.

Finally, chapter 5 will investigate a highly efficient organic bulk heterojunction

photodetector fabricated from a blend of conjugated polymer (P3HT) and small

molecule (fullerene C60). Although the organic thin film photodetector shares

fundamental photophysics with the organic photovoltaic cell in terms of the pho-

tocurrent generation process, the detector needs a slightly different approach from

that used in a photovoltaic cell. We address the effect of multilayer thin film in-

terference on the external quantum efficiency. From the numerical modelling to

calculate the optical field distribution and absorption in the layers, based on the

characterization of optical properties of individual layers comprising the detector,

we propose a bulk heterojunction photodetector without the PEDOT:PSS and LiF

6

layers that are commonly used in photovoltaic cells to achieve high short-circuit

current. Through the experimental I-V characterization, we demonstrate that it

exhibits an external quantum efficiency ηEQE = 87 ± 2% under an applied bias

voltage V = −10 V, leading to an internal quantum efficiency ηIQE ∼ 97%. These

results show that the charge collection efficiency across the intervening energy bar-

riers can indeed reach near 100% under a strong electric field.The achieved external

quantum efficiency is one of the highest values published so far.

7

Chapter 2

Optical Chiral Waveguide

2.1 Organic chiral material and optical activity

When molecules do not possess a plane of symmetry so that they are not superim-

posable on their mirror images through rotation and translation, they are referred

to as optical isomers or enantiomers and are said to be chiral. The word “chiral” is

derived from Greek meaning ‘hand’: our hands are mirror images and they cannot

be superimposed on each other no matter how hard we try.

Most of the physical and chemical properties of enantiomers are identical –

melting point, boiling point, density, solubility, etc. – but they can interact dif-

ferently in biological systems. For this reason, chiral molecules are of particular

interest in the pharmaceutical industry; of their two forms, one is clinically very

effective but the other is ineffective or even dangerous at times. One example

is thalidomide that was used for pregnant women in aiding morning sickness. It

was discovered, however, that one handedness of the molecule, (S )-thalidomide,

relieved the woman’s nausea, but the other handedness, R-enantiomer, caused hor-

rible birth defects. Another example of a chiral drug is ibupfofen, commonly found

in over-the-counter pain relievers. (R)-ibupfofen in racemic mixture is not only 100

8

times less effective, but also substantially slows the rate at which S -enantiomer

takes effect in the body [5,6].

When these chiral materials interact with electromagnetic waves, the plane of

polarization is rotated and they are said to be optically active. This optical activity

has been the subject of investigation for a few decades, especially in the microwave

and optics areas. Although quantum mechanics is required to give us a complete

understanding of the origin of the optical activity, it can simply be explained as

circular birefringence [7].

The modes in a chiral bulk material are right-handed circular polarization

(RHC) and left-handed circular polarization (LHC). When a linearly polarized

light enters into the chiral medium, it is decomposed into two orthogonal co-

propagating circular polarizations, RHC and LHC, travelling at different speeds

due to the circular birefringence. After propagating through the chiral material

and recombining, the output is linearly polarized as well, but is rotated by a certain

angle with respect to the plane of polarization of the incident wave. In general,

how much a certain chiral material rotates a plane of polarization is described

by specific rotatory power [α]25D in units of [ deg·cm2

dm·g], where 25 and D represents a

temperature of 25 C and a wavelength of the sodium D line at 589 nm, respectively.

The traditional sign convention for the optical activity in a chiral material is the

following: with light propagation through a sample towards an observer, α > 0

represents clockwise rotation and α < 0 represents counter-clockwise rotation when

viewed from the direction toward which the light is travelling (i.e. looking at the

9

light source). Sometimes, it is referred to as (+), right-handed or dextro-rotatory

chiral when α > 0, and (−), left-handed or levo-rotatory when α < 01. It should

be noted that this sign convention is exclusively used by chemists and the opposite

convention is adopted in many physics or engineering text books.

Organic chiral materials that form amorphous isotropic thin solid films have

potential use in optical waveguides for photonic applications. The recent theory [8]

on chiral-core planar waveguide suggests that an optical waveguide device may be

possible for novel applications, provided that low loss amorphous films of sufficient

chirality can be fabricated.

Useful applications include passive TE/TM mode converters for rotating the

output polarization of semiconductor lasers to optimize the polarization for down-

stream active waveguide devices in integrated optics, and active polarization con-

trol by combining an optically active waveguide segment with an electrooptic phase

shifter. There is a distinct advantage to achieving polarization rotation by optical

activity instead of linear birefringence. For a birefringent linear polarization rota-

tor, the input polarization must be at a 45 angle with respect to the optic axis

– a condition that is difficult to achieve in integrated optics. An optically active

bulk material that is isotropic, on the other hand, will rotate a linearly polarized

input of any orientation to a linearly polarized output.

To make use of this advantage in integrated optics, the eigenmodes of the chiral

1dextro and levo are from Latin meaning right and left, quite often they are used as d and l

in short, respectively.

10

waveguide must be circularly polarized. The challenge for this application is to

synthesize polymer or macromolecular structures with a high degree of chirality,

ultimately extending into the near infrared for telecommunication applications,

that also form amorphous isotropic thin films with negligible linear birefringence.

Large specific rotations on the order of 15,000 deg·cm2

dm·ghave been realized, for exam-

ple, with helicenes [9]. However, these molecules tend to aggregate in the formation

of thin films, a feature which can cause increased optical loss.

Although there has been relatively recent interest in the nonlinear optical prop-

erties of chiral materials, novel chiral photonic waveguides from linear chiral optical

effects are discussed here based on the development of chiral-core asymmetric slab

waveguides from amorphous organic binaphthyl films.

2.2 Unique polarization properties of chiral-core planar waveguides

Chirality in a dielectric planar waveguide introduces considerable mathematical

complexity, resulting in mode properties that can be significantly different from

those of the achiral case. This section reviews the theoretical analysis of the chiral

core planar (a)symmetric waveguides to provide a relatively comprehensive sum-

mary for later discussions. Although there are a number of theoretical papers

published, this section will mainly focus on reference [8] because all the analysis

performed in this chapter adopts the approach described there. It has the distinc-

tive advantage that the modal eigenvalue equations contain, in relatively simple

form, a pair of parameters determining the eccentricity of the polarization ellipse

11

for the transverse electric field.

2.2.1 Fields in bulk chiral media

When there are no “free” charges and currents inside the media (J = ρ = 0), the

electric and magnetic fields satisfy the Maxwell’s equations [10],

∇× E = −∂B

∂t, (2.1)

∇× H =∂D

∂t, (2.2)

∇ · D = 0, (2.3)

∇ · B = 0. (2.4)

The boundary conditions that must be satisfied by the electric and magnetic fields

at the interface between two different dielectric materials, designated 1 and 2, can

be expressed as

(E1 − E2) × n = 0, (2.5)

(H1 − H2) × n = 0, (2.6)

(D1 − D2) · n = 0, (2.7)

(H1 − H2) · n = 0. (2.8)

If we assume that all the field components have a time-dependence of ejωt,

E(r, t) = ReE(r) exp(jωt), (2.9)

H(r, t) = ReH(r) exp(jωt), (2.10)

12

we can rewrite the equations involving time derivatives, Eqs. 2.1 and 2.2, in terms

of complex field quantities.

∇× E = −jωB, (2.11)

∇× H = jωD. (2.12)

Next, we adopt the Drude-Born-Federov2 constitutive relation in order to introduce

material chirality from isotropic chiral media into Maxwell’s equations [13].

D = ǫ(E + γ∇× E), (2.13)

B = µ(H + γ∇× H), (2.14)

where ǫ and µ are the permittivity and permeability, and γ is the chirality having

the units of length. For reference, a chirality parameter γ of 1 pm corresponds to

a bulk rotatory power ρ of 15 deg/mm at a wavelength of 633 nm with an average

refractive index of 1.62. From Eq. 2.11–2.14, we can obtain

(1 − ω2µǫγ2)∇× E − ω2µǫγE + jωµH = 0, (2.15)

(1 − ω2µǫγ2)∇× H − ω2µǫγH − jωǫE = 0. (2.16)

These coupled equations can be reduced to Helmholtz wave equations with Bohren’s

decomposition of E and H,

F± ≡ E ± j

√

µ

ǫH, (2.17)

2Another choice for a constitutive relation describing isotropic chiral media may be the sym-

metrized Condon set, D = ǫE−jgωH and H = µH+jgωE. However, two constitutive equations,

Drude-Born-Federov and symmetrized Condon, are equivalent to first order in g and γ [11, 12].

13

or

E =1

2(F+ + F−), (2.18)

H =1

2j

√

ǫ

µ(F+ − F−). (2.19)

By using Eqs. 2.15 and 2.16, and from Eqs. 2.18 and 2.19, we can easily verify that

∇× F± = ∓k0n±F±, (2.20)

n± ≡ ng

1 ± δ, (2.21)

δ ≡ ω√

µǫγ = k0ngγ, (2.22)

where ng is the index of refraction,√

ǫ/ǫ0, since we assume µ = µ0, and k0 is the

free-space wave vector, 2π/λ0. Finally, we can obtain the Helmholtz equations for

F± by computing the curl of Eq. 2.20.

∇2F± + k20n

2±F± = 0. (2.23)

Note that, in bulk media, solutions of Eq. 2.20 have the form,

F± = (x ± jy)Foexp(−jk0n±z), (2.24)

representing plane waves with right-handed circular (RHC) and left-handed circu-

lar (LHC) polarization. In terms of handedness describing how the end point of

the electric field vector traces a circle or ellipse on a plane perpendicular to the

beam, there exist two opposing conventions. To avoid any confusion, we shall use

the convention throughout this chapter as used in most optics books: for the RHC

(LHC), the electric field vector rotates clockwise (counter-clockwise) in time at a

14

fixed position in space to an observer looking at the source of the light. In fact,

this convention seems more natural since the rotation of the electric field vector

and the direction of propagation form a right-handed screw (left-handed screw) in

space at a fixed time for the RHC (LHC) in our definition here. It is important to

note, however, that if e−jwt was used for a time dependence of field components

instead of e+jwt, the handedness would be reversed: F+ for LHC and F− for RHC.

Optical rotatory power due to the circular birefringence is given by

ρ =π

λo

(n− − n+) (2.25)

≃ k0ngδ, for δ ≪ 1, (2.26)

2.2.2 Eigenmodes in asymmetric chiral planar waveguides

As we discussed in section 2.2.1, the eigenmodes in bulk chiral media are RHC

and LHC polarizations. In a waveguide structure, however, the eigenmodes are

two orthonormal elliptically polarized modes in general – right-handed elliptical

(RHE) and left-handed elliptical (LHE) polarizations – due to influence of the

boundary conditions. This section describes in detail the polarization properties of

eigenmodes in an asymmetric chiral planar waveguide, using eigenvalue equations

in terms of a pair of parameters g and h corresponding to the ellipticity for the

transverse electric field [8]. Hence it allows a detailed insight into the polarization

properties.

Consider the planar waveguide structure shown in Fig. 2.1. The upper and

lower cladding layers are not chiral having indices of refraction no and ns, respec-

15

ng, g

x

y

0

-d

ns

no

Figure 2.1: The structure of an asymmetric planar waveguide with an isotropic

chiral core. The upper and lower cladding layers are not chiral having indices of

refraction no and ng, respectively, and the core layer has an index ng, chirality γ,

and thickness d. The coordinate axes have been oriented such that the waveguide

points in the z-direction and is assumed to be infinite in the ±x-direction.

16

tively, and the core layer has an index ng, chirality γ, and thickness d. The coor-

dinate axes have been oriented such that the waveguide points in the z-direction

and is assumed to be infinite in the ±x-direction. If we assume a solution for the

Helmholtz wave equation of the form,

F±(y, z) = Ψ±(y) exp(−jk0neffz), (2.27)

where neff is the effective index of a mode propagating along the +z-direction,

then the cartesian x-component of Ψ± satisfies the Helmholtz equation (putting

∂∂x

= 0):

dΨ±x (y)

dy2+ k2

0(n2q − n2

eff)Ψ±x = 0, (2.28)

where the subscript q in nq denotes the waveguide layer, ‘±’ in the core, ‘o’ in

the upper cladding, and ‘s’ in the lower cladding layer. As in the achiral slab

waveguide, the solution to the above equation can be written as

y =

A± exp(−vy) y ≥ 0

B± cos(u±y + φ±) 0 < y < −d

C± exp[w(y + d)] y ≤ −d

, (2.29)

where

u± ≡ k0

√

n2± − n2

eff ,

v ≡ k0

√

n2eff − n2

o, (2.30)

w ≡ k0

√

n2eff − n2

s.

17

The y and z-component of Ψ± can be obtained from Eq. 2.20 and Ψ±x (y),

Ψ±y (y) = ±j

neff

nq

Ψ±x (y), (2.31)

Ψ±z (y) = ± 1

k0nq

dΨ±x (y)

dy. (2.32)

From Eqs. 2.5 and 2.6 describing the boundary conditions that require continuity

of the tangential components of E and H across the interface at y = 0 and y = −d,

and using Eqs. 2.18–2.27, we can obtain a set of eight equations. Eliminating A±

from the four equations at y = 0, and C± from the remaining four equations at

y = −d yields

B+(σ+o sin φ+ − cos φ+) + B−(σ−

o sin φ− − cos φ−) = 0, (2.33)

B+(roσ+o sin φ+ − cos φ+) − B−(roσ

−o sin φ− − cos φ−) = 0, (2.34)

B+[σ+s sin (u+d − φ+) − cos (u+d − φ+)]

+ B−[σ−s sin (u−d − φ−) − cos (u−d − φ−)] = 0,

(2.35)

B+[rsσ+s sin (u+d − φ+) − cos (u+d − φ+)]

− B−[rsσ−s sin (u−d − φ−) − cos (u−d − φ−)] = 0,

(2.36)

where

σ±o ≡ (1 ± δ)

u±

v,

σ±s ≡ (1 ± δ)

u±

w, (2.37)

rp ≡n2

p

n2g

, p = o or s,

By introducing two equations – which equivalently satisfy the conditions that the

determinants of the coefficients B+ and B− in Eqs. 2.33 and 2.34, and Eqs. 2.35

18

and 2.36 are zeros – including new parameters g and h such that

cot φ± = σ±o

(

ro ± g

1 ± g

)

, (2.38)

cot (u±d − φ±) = σ±s

(

rs ± h

1 ± h

)

, (2.39)

and solving for the determinant, the four equations (Eqs. 2.33–2.36) can give a

single equation,

1 + h

1 − h=

S+ro

+ S+ro

g

S−ro

+ S−ro

g. (2.40)

Note that Eqs. 2.38, 2.39 and 2.40 represent five equations with five unknowns φ±,

g, h and neff , and equivalently state all the constraints imposed by the boundary

conditions. We can eliminate φ± in Eqs. 2.38 and 2.39, and solve Eq. 2.40 for h in

order to obtain eigenvalue equations along with the mode eccentricity parameters

g and h,

u±d = cot−1

(

σ±o

ro ± g

1 ± g

)

+ cot−1

(

σ±s

rs ± h

1 ± h

)

+ m±π, (2.41)

h(g, neff) =(S+

ro− S−

ro) + (S+

o − S−o )g

(S+ro

+ S−ro

) + (S+o − S−

o )g. (2.42)

It is worth noting that Eq. 2.41 has an inverse-trigonometric form, similar to the

achiral case that is familiar to us [14]. The two equations in Eq. 2.41 can be

simultaneously solved for g and neff with Eq. 2.42. For an example, Fig. 2.2 shows

the calculation for the mode effective index neff as a function of waveguide core

thickness d at a wavelength λ = 633 nm for the fundamental and the first higher

modes in an achiral (γ = 0) and a chiral (γ = 50 pm) asymmetric planar waveguide.

The refractive index of the core layer is ng = 1.62, and the upper and lower cladding

19

0 1 2 3 4 5 6 7 8 9 101.5

1.55

1.6

1.65

d (µm)

n eff

ng

TE0

TM0

TE1

TM1

(a) Achiral (γ = 0)

0 1 2 3 4 5 6 7 8 9 101.5

1.55

1.6

1.65

d (µm)

n eff

n+

n−

RHE0

LHE0

RHE1

LHE1

(b) Chiral (γ = 50pm)

Figure 2.2: Calculation of the mode effective index neff vs. waveguide thickness d,

for the fundamental and the first higher modes in (a) an achiral (γ = 0) and (b) a

chiral (γ = 50 pm) asymmetric planar waveguide. The refractive indices ng = 1.62,

no = 1.00, and ns = 1.50 are used for this calculation.

20

indices are no = 1.00 and ns = 1.50, respectively. The neff ’s for TE and TM in

the achiral case are asymptotic to the core index ng with increasing thickness far

from the cut-off. On the other hand, the neff ’s for RHE and LHE in the chiral

case are asymptotic to the core indices n+ and n−, respectively with increasing

thickness [8, 15,16].

Once g and neff are obtained, φ± can be found from Eq. 2.38, and other re-

maining constants for field amplitudes in Eq. 2.29 can be found as,

B+ = B− g + ro

g − ro

cos φ−

cos φ+, (2.43)

A± = B− cos φ− g ±√ro

g − ro

, (2.44)

C± = B− cos (u−d − φ−)h ±√

rs

h − rs

. (2.45)

From Eqs. 2.18 and 2.27, the transverse electric field is given by

ET =1

2

[

Ψ+y (y) − Ψ−

y (y)]

x + jneff

[

Ψ+y (y)

nq

−Ψ−

y (y)

nq

]

y

exp (−jk0neffz).

(2.46)

The polarization for the field given by the above equation is, in general, an elliptical

polarization with major axes on either the x or y axis. The mode eccentricity –

the ratio of the y-component to the x-component – in the upper cladding (y ≥ 0)

is given by

Ey

Ex

∣

∣

∣

∣

y≥0

= jneff

no

√ro

g= j

neff

ng

1

g, (2.47)

and the mode eccentricity in the lower cladding (y ≤ 0) is given by

Ey

Ex

∣

∣

∣

∣

y≤−d

= jneff

ns

√rs

h= j

neff

ng

1

h. (2.48)

21

The polarization ellipse in the core is more complicated than in the cladding and

varies with position in y, but considering the equation,

B+

B−=

g + 1

g − 1

√

√

√

√

√

√

1

σ+o

2 +(

g+ro

g+1

)2

1

σ−

o2 +

(

g−ro

g−1

)2 , (2.49)

gives an idea of the mode eccentricity. In limiting cases,

when γ → 0,

B− = B+, as g → ±∞ −→ TE

B− = −B+, as g → 0 −→ TM,

when γ 6= 0,

B+/B− → 1, as g → ±∞ −→ predominant TE

B+/B− → −1, as g → 0 −→ predominant TM

B+/B− ≫ 1, as g → 1 −→ predominant RHC

B+/B− ≪ 1, as g → −1 −→ predominant LHC.

Figure 2.3 illustrates the mode ellipticity parameter g as a function of chirality

γ for the fundamental elliptical modes, RHE0 and LHE0, in the asymmetric planar

waveguide with a 3µm thick chiral-core, an average core refractive index ng = 1.62,

and upper and lower cladding indices of no = 1.00 and ns = 1.50, respectively.

When γ > 0, the RHE(LHE) modes evolve from TM(TE) at γ = 0 and become

predominantly RHC(LHC) modes with increasing chirality.

22

−50 −40 −30 −20 −10 0 10 20 30 40 50

−1

1

γ (pm)

g

RHC

TM

LHC

RHE0

−50 −40 −30 −20 −10 0 10 20 30 40 50

−1

1

γ (pm)

1/g

RHC

TE

LHC

LHE0

Figure 2.3: Calculation of the mode eccentricity parameter g vs. chirality γ for the

lowest order elliptical modes, RHE0 and LHE0, in an asymmetric planar waveguide

with a 3 µm thick chiral-core, and refractive indices ng = 1.62, no = 1.00 and

ns = 1.50. When γ > 0, the RHE(LHE) modes evolve from TM(TE) at γ = 0 and

become predominantly RHC(LHC) modes with increasing chirality.

23

It should be noted that a small refractive index difference between the chiral-

core and the cladding can effectively enhance the chirality effect. In other words,

for a weakly guided waveguide structure, the mode ellipticity becomes close to

unity – more nearly circular – for a given material chirality γ.

One way to look at this quantitatively is by considering the asymptotic limit

of |σ±s | in Eq. 2.37,

|σ±s | =

∣

∣

∣

∣

∣

(1 ± δ)

√

n2± − n2

eff√

n2eff − n2

s

∣

∣

∣

∣

∣

(2.50)

≈√

2δ

1 − rs

, for neff → ng and δ ≪ 1, (2.51)

which is zero in the achiral case (δ = 0), and increases either with increasing δ

(proportional to the material chirality γ) or with increasing rs, related to the index

contrast ∆n as

∆n =ng − ns

ng

= 1 − 1√rs

. (2.52)

This can be explained qualitatively from the following: the mode in achiral

waveguides are TE and TM in general. However, the critical angle for total internal

reflection depends on the index difference between the core and cladding, therefore,

in the weakly confined waveguides, only light at a grazing incidence angle that

is bigger than the critical angle can be supported through the waveguide. This

confined mode in the achiral case can be considered as approximately TEM in

nature. Similarly, the RHE and LHE modes in chiral waveguides approximate as

RHC and LHC in bulk for the weakly confined waveguides.

It should be noted that the mode ellipticity changes dramatically in the vicinity

24

of mode crossover points – where the curve of neff for the RHE modes successively

cross over higher LHE modes, for example, neff for the RHE0 with LHE1 as in

Fig. 2.2(b). In fact, in addition to those eccentricity parameters g and h, this is one

of the most unique advantages using the theoretical developments in reference [8].

However, this does not concern us since we are primarily interested in a single

mode waveguide that supports only the fundamental modes RHE0 and LHE0.

2.3 Chiral waveguides from amorphous binaphthyl films

Planar waveguides fabricated from chiral single molecules, identified as ML-224 and

ML-226, synthesized by Prof. Green’s group at N. Y. Polytechnic University have

been investigated for detailed experimental polarization analysis. Fig. 2.4 shows

the chemical structures of both ML-224 and ML-226, which consist of binaphthyls

attached to cis-1,3,5-cyclohexanetricarboxylic ester. These binaphthyl based chi-

ral non-racemic compounds adopt a scheme designed to produce materials that

form glassy isotropic thin solid films with negligible linear birefringence [17]. The

bridged binaphthyls are based on work in [18], in which the thermal and photo-

racemization of these materials were used to explore fundamental characteristics of

the glassy state. The detailed synthetic work leading to ML-224 and ML-226 can

be found in [19]. The glass transition temperature Tg measured with a differential

scanning calorimeter (DSC) is in the range of 65−70 C and 45−50 C, and the spe-

cific rotatory power in solution measured with a polarimeter is [α]25D = +428 deg

and [α]25D = −468 deg, for ML-224 and ML-226, respectively. Note that ML-

25

X1

X2

O

O

X3

COOY

COOY

YOOC

Y =

Figure 2.4: Chemical structure of the single molecule binaphthyls attached to cis-

1,3,5-cyclohexanetricarboxylic ester. For ML-224, X1=H, X2=H and X3=CH2O-

heptyl, and for ML-226, X1=heptyl, X2=heptyl and X3=H.

224 is a right-handed (d -rotatory) chiral compound and ML-226 is a left-handed

(d -rotatory) chiral compound as shown in Fig. 2.5 for optical rotatory dispersion

(ORD) curves for both materials.

For thin solid films, the refractive index dispersion of these films was measured

using a MetriconTM prism coupler, which determines the ordinary and extraor-

dinary index of a film using two orthogonal linear polarizations at each of five

discrete laser wavelengths. Figure 2.6 shows the measured refractive indices of

the thin films fabricated from both ML-224 and ML-226, and curves fitted to the

Sellmeier dispersion relation,

n2 = A + Bλ2

λ2 − λ2o

, (2.53)

where λo is the wavelength corresponding to the absorption peak in the material.

As shown in the figure, both samples are isotropic within the experimental un-

certainty (∼ 0.0001). This isotropic property is extremely critical because even

moderate amounts of linear birefringence can overpower the effects of optical ac-

26

Wavelength (nm)

400 500 600 700 800

2000

1000

-1000

-2000

-3000

0

Sp

ec.

Ro

tatio

n

ML-224 (RH, d-rotatory)

ML-226 (LH, l-rotatory)

Figure 2.5: Optical rotatory dispersion (ORD) curves for ML-224 and ML-226.

ML-224 has a right-handed chirality and ML-226 has a left-handed chirality.

tivity.

The optical activity of the solid films was measured using a 670 nm laser with

the laser beam passing through the film at normal incidence. For these films,

normal incidence corresponds to propagation along the optic axis, and therefore

the circular birefringence can be measured independently of linear birefringence

effects. The sample films, which are several microns thick on a fused quartz sub-

strate, were placed between two Glan-Thompson polarizers in a polarizer-analyzer

configuration, and the analyzer was rotated about the null position using a com-

puter controlled precision rotation stage as depicted in Fig. 2.7. From a fit of the

data to a sine-squared curve, the change in the null position of the film on a UV-

grade fused quartz substrate was compared to the null for transmission through the

27

Inde

x o

f R

efra

ctio

n

Wavelength (nm)

Figure 2.6: Index of refraction dispersion measured with a MetriconTM prism cou-

pler for ML-224 and ML-226 in thin solid films. Both materials are isotropic within

the experimental uncertainty (∼ 0.0001).

substrate alone. Measured optical rotatory power at 670 nm is about 4−5 deg/mm

for both ML-224 and ML-226, which is consistent with the specific rotatory power

measured in solution.

Figure 2.8 depicts the structure of the asymmetric planar waveguide fabricated

from ML-224 and ML-226 for a detailed study of the polarization properties of

the eigenmodes. GaAs was used for a substrate simply due to its availability and

ease of cleaving, and SiON was selected for the lower cladding layer because its

refractive index can be easily tailored by controlling the ratio of the gas mixture in

plasma-enhanced chemical vapor deposition (PECVD). Theoretically, the nitrogen-

28

Laser DetectorPolarizer mounted on step-motorPolarizer

Sample

Power-meterStep-motor Controller

Figure 2.7: Experimental setup for measurement of optical rotatory power. The

sample films were placed between two Glan-Thompson polarizers in a polarizer-

analyzer configuration, and the analyzer was rotated about the null position using

a computer controlled precision rotation stage.

to-oxygen ratio and hence the index of SiON can be varied over a significant range

from ∼1.46, the index of silicon oxide, up to ∼2.0, the index of silicon nitride. For

the SiON lower cladding, we used an Oxford Plasmalab PECVD with a chamber

pressure of 300 mT and an RF power of 10 W at a substrate temperature of 300 C.

We fixed the flow rate of N2O and SiH4 to 12 sccm and 64 sccm, and varied the

flow rate of NH3 to reach the desired refractive index for the lower cladding, which

results in a deposition rate of 95-105 A/min.

After depositing the SiON layer on the GaAs substrate, to achieve approxi-

mately 3 µm thick films for single-mode waveguides, a solution of ML-224 and ML-

226 dissolved in cyclohexanone (∼ 27 % by weight) was spin-coated at 1000 RPM

29

GaAs (substrate)

SiON (lower cladding)

Chiral Core

Air (upper cladding)

Thickness, d ML-224 : 3.08 mm ML-226 : 2.85 mm

d

8 mm

Figure 2.8: Structure of chiral slab waveguides on GaAs substrates coated with

SiON.

ML-224 ML-226 SiON

Wavelength no ne no ne for ML-224 for ML-226

633 nm 1.6201 1.6200 1.6030 1.6030 1.6135 1.5945

Table 2.1: Refractive indices of chiral-core layers from ML-224 and ML-226, and

SiON lower cladding layers at 633 nm in the waveguide structure depicted in

Fig. 2.8.

30

for 30 s on SiON, and then cured in a Fisher Scientific vacuum convection oven

with N2 environment. The temperature was elevated from room temperature up to

Tg with a ramp rate of 3 C/min, soaked for 30 min, and naturally cooled down to

room temperature. For the final step, the waveguides were cleaved to form smooth

end facets for input and output coupling.

Two critical design points should be noted at this point. First, the waveguide

thickness was determined in order to satisfy single mode operation supporting the

first two orthogonal modes RHE0 and LHE0 only. Figure 2.9 shows the calculation

of mode effective index with increasing core thickness for each waveguide. The cut-

off thickness of the first higher order elliptical mode is 3.15 µm for the waveguide

from ML-224 and 2.8 µm for the waveguide from ML-226. Second, the refractive

index of the SiON lower cladding layer was adjusted to be close to that of the chiral-

core layer. As pointed out in section 2.2.2, a small index difference between the

chiral-core and the cladding layer, ng and ns, respectively, is desirable because it

can enhance the eccentricity of elliptical eigenmodes for a given amount of material

chirality. The index contrast, ∆n = (ng − ns)/ng at a wavelength of 633nm, was

about 0.004-0.005 as shown in Tab. 2.1. Figure 2.10 shows the calculation of

ellipticity of the fundamental modes, RHE0 and LHE0 as a function of chirality γ

for three different lower cladding indices, 1.613, 1.550 and 1.500. When γ = 1.0 pm,

for example, the eccentricity parameter g almost doubles when ns increases from

1.500 to 1.613.

31

0 2 4 6 8 10

1.614

1.616

1.618

1.620

dc(LHE

1) = 3.15 µm

ML-224; ng=1.620, n

s=1.6135, γ =+0.3pm

mode e

ffective index, n

eff

waveguide thickness, d (µm)

LHE1

RHE1

RHE0

LHE0

0 2 4 6 8 10

1.596

1.598

1.600

1.602

dc(RHE

1) = 2.80 µm

ML-226; ng=1.603, n

s=1.5945, γ =-0.3pm

mode e

ffective index, n

eff

waveguide thickness, d (µm)

LHE1

RHE1

RHE0

LHE0

Figure 2.9: Calculation of neff with increasing waveguide core thickness. (a) The

cut-off thickness of LHE1 mode is about 3.15 µm for the waveguide from ML-224.

(b) The cut-off thickness of RHE1 mode is 2.8 µm for the waveguide from ML-226.

32

0 5 10 15 20-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

ng=1.620, d =3.08 µm

g (

∼ E

x/E

y )

cladding index (ns)

1.613

1.550

1.500

LHC

RHC

chirality, γ (pm)

1/g

(

∼ E

y/E

x )

Figure 2.10: Calculation of mode ellipticity of RHE0 and LHE0 of asymmetric slab

waveguide as a function of given material chirality γ for the core index ng = 1.620

and three different lower cladding indices, ns = 1.613, 1.550, and 1.500. When

a γ = 1.0 pm, for example, the eccentricity parameter g almost doubles when ns

increases from 1.500 to 1.613.

33

2.4 Experimental polarization analysis

In previous sections we discussed the theoretical polarization properties of the

eigenmodes and fabrication of asymmetric planar chiral-core waveguide from bi-

naphthyl based chiral compounds including key design parameters. For experi-

mental polarization analysis of these thin film waveguides, the state of polarization

(SOP) represented by normalized Stokes parameters of the eigenmodes was charac-

terized by determining the SOP at the output of the waveguide for two orthogonal

input polarizations: horizontal and vertical polarizations. We refer to these or-

thogonal linearly polarized inputs as TE and TM with respect to the slab surface

because, in an achiral slab waveguide, they would excite usual TE and TM modes.

For chiral waveguides, a linearly polarized input will excite both eigenmodes, RHE

and LHE, and the TE input excites a different relative fraction of these elliptical

eigenmodes than the TM input. As the light propagates down the waveguide, the

RHE and LHE modes travel at different speeds resulting in an overall elliptical

polarization with a rotated major axis at the output of the waveguide. If the

eigenmodes are circular, the output polarization is a linear polarization as well

but rotated by a certain angle. By determining the output polarization state for

a given input state, the eigenmodes of the slab waveguide can be computed.

As illustrated in Fig. 2.11 for the experimental setup, we used an intensity

stabilized He-Ne laser for a light source and a Faraday isolator after the laser

to prevent any undesired back-reflection into the laser cavity from deteriorating

34

Intensity-stabilized He-Ne Laser

Faraday

Isolator

Polarization Controller

Objective

LensCleaved Fiber

Single-mode Fiber

Collimator

Aperture

Power-meter

Step-motor Controller

Detector

Polarizer

Rotatingλ/4 plate

Figure 2.11: Experimental setup for polarization analysis for chiral waveguide.

the laser stabilization. Light is then directed through a polarization controller to

launch a specific polarization into the waveguide under measurement. An objective

lens was used for input coupling and a cleaved single mode fiber was used for

output coupling, both mounted on a 3-axis translation stage, allowing accurate

and stable alignment. The light signal travels through the additional single mode

fiber after the waveguide, and it is collimated and detected after passing through a

polarization analyzer section consisting of a rotating quarter waveplate and a linear

polarizer. The quarter waveplate is mounted on a rotating step motor controlled

by a computer.

The polarization analyzer section in Fig. 2.11 can be described using Stokes

35

vector and Mueller matrix representation. We can write a relationship,

S′ = M · S

= LP(0) · QWP(θ) · S

=

12

12

0 0

12

12

0 0

0 0 0 0

0 0 0 0

1 0 0 0

0 cos2 2θ sin 2θ cos 2θ − sin 2θ

0 sin 2θ cos 2θ sin2 2θ cos 2θ

0 sin 2θ − cos 2θ 0

Sin0

Sin1

Sin2

Sin3

, (2.54)

where S is the Stokes vector describing the SOP we want to determine, and LP(0)

and QWP(θ) are the 4× 4 Mueller matrices representing a fixed horizontal linear

polarizer and a rotating quarter waveplate with an azimuthal angle θ with respect

to the horizontal axis, respectively. From Eq. 2.54, the S ′0-component in S′ =

(S ′0, S

′1, S

′2, S

′3) is given by

S ′0(θ) =

1

2· S0 +

1

2cos2 2θ · S1 +

1

4cos 4θ · S2 −

1

2cos 2θ · S3, (2.55)

which is a function of the Stokes parameters S0, S1, S2 and S3 in S that we want

to determine, and it is equivalent to the total intensity Io measured by a detector,

assuming the detector is polarization insensitive. Once we measure the intensity

at varying angles of θ, the unknown SOP represented by S can be obtained from

linear regression.

We should recognize, however, that the Mueller matrix of the fiber used for

output coupling must be accounted for in order to obtain the SOP at the output

of the waveguide (or equivalently at the input of the fiber). This can be done

36

by a calibration process in which three different input polarizations, [s1, s2, s3]3 =

[1 0 0], [0 1 0] and [0 0 1] representing horizontal linear polarization, 45 linear

polarization and RHC polarization, respectively, are launched into the fiber and

the resulting output SOP for each case is determined from the linear regression as

above [20]. This calibration procedure finds a characteristic 3× 3 Mueller matrix4

Mfiber [21] that determines the SOP at the output of the fiber for a given input

state. The matrix Mfiber can then be inverted to determine the SOP at the input

of the fiber (and thus the output of the chiral waveguide) when the SOP at the

output of the fiber is known.

Figure 2.12 shows the measurement results to determine the Stokes parame-

ters corresponding to the polarization states of the waveguide modes after the

additional single mode fiber. The dots represent the detected optical power as a

function of θ, the angle of the fast axis of a rotating quarter waveplate with respect

to the transmission axis of a fixed linear polarizer for two input polarizations sI of

TE and TM. As discussed above, from linear curve fitting and after taking into

account Mfiber, we obtained polarization states at the waveguide output sF

3[s1, s2, s3] is a normalized Stokes vector obtained by dividing the 4-Stokes vector by the total

optical power S0, i.e., s1 = S1/S0, s2 = S2/S0, s3 = S3/S0.

4When we ignore the loss in the fiber, it can be considered as a unitary optical system and

represented by the 3 × 3 Mueller matrix, a subset of a 4 × 4 Mueller matrix in which the first

row and the first column is always [1000] and [1000]T for the unitary optical component.

Mfiber =

m11 m12 m13

m21 m22 m23

m31 m32 m33

37

0 50 100 150 200 250 300 3500

0.2

0.4

0.6

0.8

1

Angle (Deg)

Nor

mal

ized

Inte

nsity

TE input, measuredTE input, fittedTM input, measuredTM input, fitted

(a) ML-224

0 50 100 150 200 250 300 3500

0.2

0.4

0.6

0.8

1

Angle (Deg)

Nor

mal

ized

Inte

nsity

TE input, measuredTE input, fittedTM input, measuredTM input, fitted

(b) ML-226

Figure 2.12: Detected optical power as a function of the angle between the rotating

quarter waveplate and polarizer to determine the Stokes parameters corresponding

to the polarization states of the waveguide modes. The solid lines fit to Eq. 2.55

for slab waveguides fabricated from (a) ML-224 and (b) ML-226.

38

for ML-224 waveguide

sFTE input =

0.75

−0.42

0.50

, sFTM input =

−0.72

0.45

−0.51

, (2.56)

and for ML-226 waveguide

sFTE input =

0.55

−0.06

−0.81

, sFTM input =

−0.53

0.08

−0.82

. (2.57)

Next, we want to obtain the Stokes parameters describing the eigenstates of the

waveguide from the SOP at the waveguide output for a given input polarization.

The continuous evolution of the polarization states through the waveguide can be

described by

dS

dz= Ω × S, (2.58)

where Ω is the waveguide eigenmode with projections Ω1, Ω2, and Ω3 defined in

the same coordinate space as the Stokes vector S. The above equation states that

while the light signal propagates down the waveguide from any initial position,

the evolution of the polarization state continuously traces out a circular orbit

perpendicular to an axis connecting orthogonal eigenstates on the Poincare sphere.

The evolution from the initial state sI to the final state sF can be easily visualized

on the Poincare sphere as illustrated in Fig. 2.13 [22]. For example, when the

eigenstates are circularly polarized modes, in other words, when the vector Ω is

39

Ω

SISF

S1

S3

S2

(a) Circularly polarized eigenstates

Ω

SI

SF

S1

S3

S2

Θ

(b) Elliptically polarized eigenstates

Figure 2.13: Continuous evolution of polarization states traces out a circular orbit

perpendicular to an axis connecting orthogonal eigenstates on the Poincare sphere.

(a) For circularly polarized eigenstates, the initial TE input polarization moves

along the equator and remains a linear polarization rotating continuously. (b) For

elliptically polarized eigenstates, the initial TE input polarization evolves into an

elliptical polarization in general.

40

directed along the S3 axis as shown in Fig. 2.4(a), the initial TE (s = [1 0 0]) input

polarization moves along the equator and remains a linear polarization rotating

continuously. If the initial polarization state is represented by a point either at

the north pole (RHC, sI = [0 0 1]) or at the south pole (LHC, sI = [0 0 − 1]),

the SOP does not change. On the other hand, when eigenstates are elliptically

polarized modes as shown in Fig. 2.4(b), the initial TE input polarization becomes

an elliptical polarization in general. As discussed in section 2.2.2, the eigenmodes

for a chiral planar waveguide are elliptical polarizations in general with major axes

on either the x or y axis, which correspond to the points on the S1 − S3 meridian

plane. So the eigenvector Ω can be written as

Ω =

sin Φ

0

cos Φ

, (2.59)

where Φ is the angle between the S3 axis and Ω as depicted in Fig. 2.13. In terms

of two angles Φ and Θ, the relationship between sI and sF can be written as

sF = Rs2(−Φ) · Rs3(Θ) · Rs2(Φ) · sI (2.60)

where RS2and RS3

are the matrices for rotational vector transformation with

41

respect to the S2 and S3 axis, respectively, and they are given by

RS2(Φ) =

cos Φ 0 − sin Φ

1 1 0

sin Φ 0 cos Φ

, (2.61)

RS3(Θ) =

cos Θ sin Θ 0

− sin Θ cos Θ 0

0 0 1

. (2.62)

For the TE input polarization (sI=[1 0 0]), after some algebraic manipulation with

trigonometric identities, we get

sF1 = 1 − 2 cos2 Φ sin2 Θ

2,

sF2 = 2 cos Φ sin

Θ

2cos

Θ

2, (2.63)

sF3 = 2 cos Φ sin Φ sin2 Θ

2,

and we can compute the angles ΦTE and ΘTE,

ΦTE = tan−1

(

sF3

1 − sF1

)

, (2.64)

ΘTE = 2 tan−1

(

sF3

sF2 sin ΦTE

)

. (2.65)

Similarly for the TM input polarization (sI=[−1 0 0]),

ΦTM = tan−1

( −sF3

1 + sF1

)

, (2.66)

ΘTM = 2 tan−1

(

sF3

sF2 sin ΦTM

.

)

(2.67)

42

For ML-224 waveguide

ΩTE input = ±

0.88

0

0.47

, ΩTM input = ±

−0.90

0

−0.44

, (2.68)

and for the ML-226 waveguide,

ΩTE input = ±

0.88

0

−0.48

, ΩTM input = ±

−0.88

0

0.47

, (2.69)

Since s3 for the polarization ellipse with major axes on either the x or y axis

is given by

s3 =2ExEy

E2x + E2

y

, (2.70)

it can be related to the ellipticity g when neff ≈ ng as in

s3 ≈2g

1 + g2for neff ≈ ng. (2.71)

From the above relation, the ellipticity parameter g is approximately 0.25 for the

waveguides from both ML-224 and ML-226. Figure 2.14 illustrates the evolution

of the SOP through the chiral waveguides fabricated from ML-224 and ML-226.

In order to look at the quantitative property of how the SOP evolves along the

waveguide as a function of the propagation distance z and the difference in mode

effective indices nLHE0

eff − nRHE0

eff , we can multiply a scale factor k0(nLHE0

eff − nRHE0

eff )

to the Ω in Eq. 2.58. However, often an alternative approach using the Jones

representation can be more intuitive and useful. From two orthonormal Jones

43

S3

S2

S1

SI = TE

0.75

-0.42

0.50(SF = )

-0.72

0.45

-0.51(SF = )

ΩRHE

(a) ML-224

S3

S2

S1

SI = TE

ΩLHE

0.55

-0.06

-0.81(SF = )

-0.53

0.08

0.82(SF = )

(b) ML-226

Figure 2.14: Evolution of polarization through the chiral waveguides from (a) ML-

224 and (b) ML-226.

44

vectors for the basis vectors corresponding to elliptical polarization states, which

are given by

E = Err + Ell, (2.72)

r =1√

a2 + b2

a

jb

, (2.73)

l =1√

a2 + b2

b

−ja

, (2.74)

r† · r = l† · l = 1, r† · l = l† · r = 0, (2.75)

we can construct a unitary transformation matrix representing a change of basis

states from Cartesian to elliptical

U =1√

a2 + b2

a −jb

b ja

, (2.76)

satisfying

Er

El

= U

Ex

Ey

, (2.77)

Now for the Jones vector after propagating a distance d through the chiral

waveguide with linear basis, we can write

E ′x

E ′y

z=d

= U† · Mwg · U ·

Ex

Ey

z=0

, (2.78)

where

Mwg =

e−jk0nrd 0

0 e−jk0nld

. (2.79)

45

If we put [Ex Ey] = [1 0], nr = nRHE0

eff , and nl = nLHE0

eff into Eq. 2.78, we can

achieve the polarization state [E ′x E ′

y] at z = d evolved from TE input at z = 0.

Note that the angle Θ in Fig. 2.4 is related to the propagation distance z and

∆neff = nLHE0

eff − nRHE0

eff such that

Θ = 2ρeffz, (2.80)

ρeff =π

λ(nLHE0

eff − nRHE0

eff ). (2.81)

Therefore, ρeffz = π corresponds to one complete circle around the eigenstate Ω.

The measured waveguide lengths d are about 16.5 mm and 11.7 mm, and from the

numerical calculations as shown in Fig. 2.9, the mode effective index differences are

approximately neff = 5 × 10−5 and ∆neff = −8 × 10−5 for the ML-224 and ML-

226 waveguides, respectively. Once these values are plugged into Eq. 2.80, we can

notice that, for both waveguides, the polarization states evolve in such a way that

they reach sF as given in Eqs. 2.56 and 2.57 after making one full complete circle.

The experimental results show that the eigenmodes of the chiral waveguides are

indeed RHE and LHE, and they are in excellent agreement with what is expected

from the theory in [8] in terms of the mode ellipticity and the SOP evolution

through the chiral-core planar waveguides.

Nevertheless, these Ω3 values are not large enough for these materials to be

used for a polarization controller or a TE/TM converter. We discussed that the

Ω3 of the eigenmode must be very close to unity, corresponding to predominantly

right- and left-handed circularly polarized waveguide modes. This can be achieved

46

0 5 10 15 20-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

ng=1.620, d =3.08 µm

1.613/1.62/1.613, Symmetric

LHC

chirality, γ (pm)

g (

∼ E

x/E

y )

1/g

(

∼ E

y/E

x )

RHC

1/1.62/1.613, Asymmetric

Figure 2.15: Calculation of mode ellipticity of RHE0 and LHE0 as a function of

given material chirality γ for the symmetric and asymmetric chiral-core planar

waveguides. The dotted line is associated with an asymmetric structure with

refractive indices ng = 1.62, n0 = 1.00 and ns = 1.613, and the solid line with a

symmetric structure with indices ng = 1.62 and n0 = ns = 1.613.

principally from larger specific rotatory power (or equivalently larger chirality γ)

of the material itself.

Figure 2.15 shows a calculation of the mode ellipticity of RHE0 and LHE0 as

a function of given material chirality γ for the symmetric and asymmetric chiral-

core planar waveguides. The dotted line is associated with an asymmetric structure

with refractive indices ng = 1.62, n0 = 1.00 and ns = 1.613, and the solid line with

a symmetric structure with indices ng = 1.62 and n0 = ns = 1.613. From the plot,

47

it is evident that, in order to have predominantly circular eigen-polarizations, a

chirality γ > 20 pm, or equivalently ρ > 300 deg/mm is required in the asymmetric

waveguide. On the other hand, in the symmetric waveguide, near circularly polar-

ized eigenstates can be obtained with considerably a lower chirality γ ≈ 4− 5 pm,

or equivalently ρ = 45 − 60 deg/mm.

The desire for circularly polarized modes, rather than just elliptically polarized

modes, is driven by the goal of achieving a pure polarization rotator in a waveguide,

i.e. a device for which a linearly polarized input is rotated to some other linearly

polarized state for any propagation length. Because even moderate amounts of

linear birefringence can overpower the effects of optical activity, an essential re-

quirement that must be fulfilled in order to attain this goal is the synthesis of a

chiral material that is isotropic in a solid film.

2.5 Other chiral materials

Other than the chiral single molecules, ML-224 and ML-226, we have charac-

terized several organic chiral materials from strong collaborations with university

chemists: Prof. Mark Green’s group at New York Polytechnic University, Prof. An-

drzej Rajca’s group at the University of Nebraska, and Prof. Lin Pu’s group at the

University of Virginia. Although the improvement in material chirality was modest

so that they were not realized into actual devices, here we summarize the charac-

teristics of those materials including optical rotatory power and index dispersion

measured using the same method described in section 2.4.

48

O O

OR

RO

O O

OR

RO

Figure 2.16: Chemical structure of the main chain binaphthyl polymer, Zhang-IV-

37 with R = n-C6H13.

2.5.1 Binaphthyl polymer

A main chain binaphthyl polymer identified as Zhang-IV-37 shown in Fig. 2.16

was supplied by the University of Virginia. The glass transition temperature Tg is

expected to be in the range of 200 C and the specific rotatory power measured in

solution is [α]25D = −711 .

For characterization, a 0.95 µm thick sample was prepared on a UV-grade fused

quartz by dissolving the solid powder in cyclohexanone to make a 10 % solution

and spin-casting followed by curing in an oven at 100 C for two hours. The optical

rotatory power measured is −8.7 deg/mm at 670 nm, which is almost twice as big

as ML-224 and ML-226. However, it exhibited a marked linear birefringence of

49

400 600 800 1000 1200 1400 16001.60

1.62

1.64

1.66

1.68

1.70

Index o

f re

fraction

Wavelength (nm)

no

ne

Figure 2.17: Index of refraction dispersion measured with a MetriconTM prism

coupler for the polybinaphthyl Zhang-IV-37 in thin solid films, which exhibited a

marked large linear birefringence, ∼ −0.01.

∼ −0.01 as shown in Fig. 2.17.

From Eq. 2.21 with Eq. 2.22, we can notice that this appreciably large linear

birefringence is almost 300 times bigger than the circular birefringence, and hence

the eigenstates of the planar waveguide would be approximately TE/TM modes.

2.5.2 Polymethine dyes

A chiral polymethine dye identified as HT-dye14 was supplied by NY Polytechnic

University and its polymer composite with a poly(butyl methacrylate-co-methyl

methacrylate) was characterized. The glass transition temperature Tg for HT-

dye14 and poly(butyl methacrylate-co-methyl methacylate) are about 122 C and

50

200 400 600 800 1000 1200 1400 1600 1800 20000

2

4

6

8

Abs. coeffic

ient (µ

m-1)

Wavelength (nm)

HT-dye14 100%

HT-dye14 22%

HT-dye14 3%

Figure 2.18: UV-Vis-IR spectra measured with a Cary 5000 spectrophotometer for

HT-dye14 in poly(butyl methacrylate-co-methyl methacrylate) copolymer.

64 C, respectively. Due to the π−π∗ absorption of the chromophore in polymethine

dyes, there is a strong absorption in the 550–700 nm spectral range. Figure 2.18

shows UV-Vis-IR spectra for the absorption coefficient measured with a Cary 5000

spectrophotometer for three samples with different concentrations of HT-dye14 in

the copolymer.

Both materials were dissolved separately in chloroform and mixed afterwards.

After casting the film, it was cured at a temperature of 170 C, substantially

above the Tg, for an hour because annealing can be quite effective in removing

51

birefringence arising from strain in the evaporation process [23]. Optical rotatory

power measured at 751 nm was about -3 deg/mm for the sample consisting of 22 %

dye in the copolymer, and -72 deg/mm for the sample from pure HT-dye14 without

the copolymer. The linear birefringence of pure HT-dye14 is 0.004−0.01 depending

upon wavelength. As we decrease the concentration of polymethine dye molecules,

the material composite becomes isotropic. For example, the sample consisting

of 22 % dye was isotropic within experimental uncertainty, ∼ 0.0001, but with a

trade-off in optical activity, -3 deg/mm.

2.5.3 (7)Helicene

Thiophene-based carbon-sulfur (7)helicene chiral molecular glass was supplied by

the University of Nebraska [24]. Tg is about 67 C and the specific rotatory power

measured in solution is [α]25D = +1166 . A 5.6 µm thick sample was made by drop-

dispensing from the solution in cyclohexanone. The optical rotatory powers were

11 deg/mm and 6 deg/mm at wavelengths of 670 nm and 850 nm, respectively, and

the linear birefringence was ∼ 0.0003.

2.5.4 Bridged binaphthyl with chromophore

Bridged binaphthyls with chromophore groups attached, identified as HT-386, was

supplied by NY Polytechnic University. A 2 µm thick sample was prepared by

spin-casting the solution in cyclohexanone. The optical rotatory power measured

was about 11 deg/mm at 670 nm. The refractive index measurement shows the

sample is isotropic within the experimental uncertainty (∼ 0.0001). However, Tg

52

is around 25 C, so it was difficult to fabricate waveguide structures.

2.6 Summary

In this chapter, we discussed the unique polarization properties of optical chiral

waveguides that can be used for integrated-optic polarization converters or polar-

ization modulators provided that low loss amorphous films of sufficient chirality

can be fabricated. Relatively recently, new modeling results for (a)symmetric

planar waveguides with chiral cores showed that eigenmodes are elliptically po-

larized modes, in general, and the polarization ellipticity depends on the chirality

that is related to the bulk rotatory power of material. For an important design

consideration, dependence of mode ellipticity on cladding index was discussed.

Through a detailed experimental polarization analysis on asymmetric chiral-core

planar waveguides fabricated from binaphthyl-based organic single molecules with

a layer of SiON for the cladding layer, we showed that eigenmodes in planar chiral-

core waveguides are indeed two orthonormal elliptical polarizations, and achieved

waveguide modes with a polarization eccentricity of 0.25, which agrees very well

with the theory. Although the achieved mode ellipticity is not large enough to

meet the requirement – predominant circularly polarized modes – for practical

applications, this is, to the best of our knowledge, the first experimental demon-

stration of the mode ellipticities for the transverse electric field of the chiral-core

optical waveguides. As we discussed, even in the case of weekly confined sym-

metric waveguide structure, the chirality γ > 5pm is required for near circularly

53

polarized eigenmodes. Characterizations and feasibilities of several other chiral

materials were also discussed. Unfortunately, none of these materials exhibited

both the appreciable material chirality and negligible linear birefringence at the

same time. The design and synthesis of such an isotropic material with a high

degree of chirality, which also extends the ORD maxima into the near infrared, is

in progress.

54

Chapter 3

Organic Magneto-Optic Material

3.1 Magneto-optic effect

The magneto-optic effect refers to a phenomenon in which light propagating through

a medium experiences birefringence in the presence of a magnetic field. There

are two well-known magneto-optic effects, distinguished by the direction of the

magnetic field with respect to the direction of light propagation, Faraday and

Cotton-Mouton. The Cotton-Mouton effect arises from a linear birefringence pro-

portional to the square of the transverse magnetic field flux density. The magnetic

field-induced linear birefringence ∆n can be written as

∆n = n‖ − n⊥

= CCMλB2t , (3.1)

where n‖ and n⊥ are the refractive indices for light polarized parallel and perpen-

dicular to the magnetic field, respectively, CCM is the Cotton-Mouton coefficient,

and Bt is the transverse component of the magnetic field. Therefore, the phase

55

difference ∆φCM between the two orthogonal polarizations is given by

∆φCM =2π

λ∆nl

= 2πCCMB2t l, (3.2)

where l is the effective optical path length through a medium.

On the other hand, the Faraday effect comes from a circular birefringence

linearly proportional to the axial component of the magnetic field. Between these

two, the Faraday effect is well studied because of its practical importance for

useful optical components, such as isolators, circulators, polarization modulators,

and sensors including current sensors and magnetic field sensors.

The Faraday effect, the magnetic field-induced optical activity, was discovered

in 1845 by M. Faraday. The plane of polarized light can be rotated by allowing

light to propagate through a medium with inversion symmetry in the presence of

a magnetic field. The angle of rotation is given by

φF = V

∫

B · dl, (3.3)

where B is the magnetic flux density, dl is an infinitesimal displacement vector

along the direction of propagation of light, and V is a proportional constant called

Verdet constant. In words, the amount of polarization rotation is proportional to

the strength of the longitudinal component of the magnetic field and the length of

the medium the light propagates through.

The Verdet constant is associated with the specific material and in general

varies with wavelength and temperature [25–27]. The units of V are often ex-

56

pressed in [rad/T · m] or [min/G · cm]. By convention, a positive Verdet constant

corresponds to a counter-clockwise rotation when light propagates parallel to an

applied B-field and viewed from the direction toward which light is travelling (i.e.

looking at the light source) [28,29]. In other words, for V > 0, the plane of polar-

ization rotates in the same direction as the current through a solenoid coil around

the medium, regardless of the direction of light propagation. Note that this con-

vention for the sense of rotation is opposite from the optical activity in a chiral

medium, in which α > 0 represents a clockwise rotation when looking at the light

source as discussed in the previous chapter.

From the consideration of the reciprocity nature in optics, there is an important

difference between the Faraday effect and optical chiral activity. In the Faraday

effect, the rotation of polarization is non-reciprocal because the interaction of light

with matter under a magnetic field breaks the time reversal symmetry and the

parity conservation [30]. A detailed theoretical discussion on the origin of the

magneto-optic effect from a quantum mechanical approach is beyond the scope of

this thesis and will be omitted.

Most recently, a very high Verdet constant from an organic polymer thin film

has been reported [31]. Although there is a lack of theoretical understanding on the

physics of the high Verdet constants in organic polymers at this point, such kinds of

materials are essential to constructing integrated-optic devices such as isolators and

circulators from simple fabrication method of spin-casting and photolithography.

57

3.2 Balanced homodyne detection

To measure the Verdet constant associated with a particular material, we can

simply increase an effective path length or use a strong magnetic field to produce

a rotation angle large enough until it becomes detectable. However, in order to

detect a small angle of rotation from a thin organic film with a typical thickness of

10 − 100 µm in a moderate magnetic field < 100 Gauss, a measurement technique

with high precision and improved signal-to-noise ratio is essential. One of the

simplest ways to measure the Faraday rotation is based on the polarizer-analyzer

configuration that we used to measure the optical rotatory power of a chiral thin

film, but it suffers from poor measurement sensitivity and accuracy, on the order of

a few mdeg. The method relies highly on the accuracy and repeatability of a step

motor used in the measurement. A few other schemes for sensitive measurement

have been reported [32, 33]. However, the sensitivity is generally worse than a

balanced homodyne detection that we adopted in our experiment [34,35].

In optical interferometry, in order to detect a phase or amplitude change in an

optical signal, the signal and reference are mixed for phase comparison. When the

reference frequency ωLO from a local oscillator is slightly different from the signal

frequency ωS, the average current detected arises from an optical field at the differ-

ence frequency ωLO − ωS (often referred to as the intermediate frequency, IF), but

no other frequencies such as 2ωLO or 2ωS are detected simply because semiconduc-

tor detectors do not respond rapidly enough. The advantage of this Heterodyne

58

detection is that one can increase the S/N ratio by increasing the local oscillator

power and achieve shot-noise limit detection in which shot noise dominates over

all other noise contributions. On the other hand, when the frequency of the local

oscillator is same as that of the signal (ωLO = ωS), but there is a relative phase

difference, it is called Homodyne detection. In homodyne detection, the phase dif-

ference is transformed into an intensity change after an optical mixer, therefore the

total intensity at the photodetector is determined by the relative phase difference

between the signal and the local oscillator [36,37].

In a balanced homodyne interferometry scheme designed to measure the Verdet

constant with a high sensitivity and S/N ratio, a linearly polarized light oriented at

45 can be used as a probe beam instead of using two separate beams. Therefore,

the two orthogonal principal polarizations – horizontal and vertical – can be treated

as the two separate beams for the signal and the reference. Once the plane of

polarization is rotated by a small angle after propagating through the magneto-

optic sample placed in a magnetic field, the small amplitude changes are projected

onto the corresponding principal components of a polarizing beam splitter as shown

in Fig. 3.1.

In addition, if a sample is placed in an AC magnetic field, the rotation of polar-

ization from an alternating magnetic flux density is transformed into an amplitude

modulation along the principal axis after passing through the polarizing beam

splitter. Because the amplitude changes along the principal axes are 180 out of

phase from each other, by using a balanced photodetector consisting of two pho-

59

Polarizer Sample PBS

I1

I2

x

y

45o

x

y

45o

Figure 3.1: Schematics of balanced homodyne detection.

−40 −30 −20 −10 0 10 20 30 40−1

−0.5

0

0.5

1

Angle of rotation (deg)

Nor

mal

ized

inte

nsity

I1

I2

I1−I

2

Figure 3.2: Normalized intensities along the principal axes I1, I2, and difference I1−I2. The amplitude changes along the principal axes from rotation of polarization

are 180 out of phase from each other.

60

todiodes and a differential amplifier, and a lock-in amplifier, we can minimize the

amplitude noise of the optical field and maximize the sensitivity with a significant

rejection of common-mode noise.

We can use the Jones formalism for a quantitative analysis of the balanced

homodyne detection. The electric field E1 and E2 at each detector can be written

as

E1,2x

E1,2y

= Pol(θ1,2p ) · HWP(θh) · Rot(φ) · 1√

2

1

1

, (3.4)

where Pol(θ1,2p ), HWP(θh), and Rot(φ) represent the Jones matrices for the po-

larizing beam splitter, the half-wave plate with the fast-axis oriented at θh, and

the rotation by an angle of φ, respectively, which are given as

Pol(θ1,2p ) =

cos2 θ1,2p sin θ1,2

p cos θ1,2p

cos θ1,2p sin θ1,2

p sin2 θ1,2p

, (3.5)

HWP(θh) =

j cos2 θh − j sin2 θh j sin 2θh

j sin 2θh j sin2 θh − j cos2 θh

, (3.6)

Rot(φ) =

cos φ − sin φ

sin φ cos φ

. (3.7)

Because the polarizing beam splitter is aligned in such a way that θ1p = 0 and

θ2p = π/2, and each detector measures the intensity of the corresponding principal

component (horizontal and vertical polarization), the electric fields E1 and E2 at

61

each detector from Eq. 3.4 can be written as

E1x

E1y

=1√2

sin (2θh − φ) + cos (2θh − φ)

0

, (3.8)

E2x

E2y

=1√2

sin (2θh − φ) − cos (2θh − φ)

0

, (3.9)

and the output from the balanced detector can be represented as

I ∝ I1 − I2

∝ E†1 · E1 − E

†2 · E2 (3.10)

∝ sin 2(2θh − φ) + constant (3.11)

= A sin 2(2θh − φ) + C. (3.12)

When an alternating magnetic field is applied, the modulated intensity at a

fixed bias point θh = θh0 is given by

∆I = −2A cos 2(2θh0 − φ)∆φ, (3.13)

∆φ = V lB0 sin ωt, (3.14)

φF = V lB0. (3.15)

Eq. 3.12 can be referred to as an optical bias curve or a transfer curve, which can

be obtained from measuring the output from the balanced detector while rotating

the half-wave plate without a magnetic field applied. Note that obtaining the

optical bias curve is a necessary step for the absolute measurement of a rotation

angle from the corresponding amplitude modulation. In other words, once we

62

obtain the A and C in the optical bias curve from curve fitting, we can convert

the demodulated signal ∆I from the lock-in amplifier into the equivalent rotation

angle of plane polarization.

Also note that the bias point θh0 can be any arbitrary value for measuring

the Faraday rotation, but the maximum demodulated signal |∆I| from the lock-in

amplifier can be achieved at 2θh0 = φ. This is the case of 45 input polarization

and horizontal alignment of the fast axis of the half-wave plate. Even when there

is a misalignment of the input polarizer or inherent linear birefringence in the

sample, the bias point for maximum demodulation can be found from the optical

bias curve.

3.3 Measurement of Verdet constant

In this section, we discuss the experimental measurement of the Verdet constant

in detail and present the measurement result for terbium gallium garnet (TGG),

one of the most popular Faraday materials used for free space optical isolators,

and for the organic samples.

Figure 3.3 depicts the experimental setup we used to measure the Verdet con-

stant. A Glan-Thompson polarizer with an extinction ratio better than 50 dB was

used to produce a linearly polarized input oriented at 45 . The sample was placed

in a Helmholtz coil through which a sinusoidal current was applied from a power

supply controlled by a function generator. We used a KEPCO BOP70-12 power

supply to amplify the voltage signal from a function generator to provide suffi-

63

Laser

Balanceddetector

Polarizer

Sample

Function generator

Helmholtz coil

Lock-in

Power supply

Wollastonprism

λ/2

Step-motor

Controller

Figure 3.3: Experimental setup of balanced homodyne detection to measure the

Verdet constant. The rotation of polarization through the sample placed in an

alternating magnetic field is transformed into an amplitude modulation along the

principal axis after passing through the Wollaston prism, and detected by a bal-

anced detector.

64

0 2 4 6 8 10 120

50

100

150

200

250

Vpp (V)

Mag

netic

Fie

ld (

G)

10 Hz

50 Hz

100 Hz

200 Hz

Figure 3.4: Calibration of the amplitude of alternating magnetic flux density inside

the Helmholtz coil for a given input AC voltage signal to power supply. Due to

the inductive impedance, the magnetic field decreases with increasing frequency.

cient current through the Helmholtz coil. After light passes through the Wollaston

prism, the rotation of polarization from an alternating magnetic flux density is

transformed into an amplitude modulation along the principal axis (horizontal and

vertical), and detected by a New Focus balanced detector and a Signal Recovery

lock-in amplifier.

As a first step, an alternating magnetic field flux density B0 sin ωt inside the

Helmholtz coil for a given AC signal was calibrated using a gauss-meter from T.W.

Bell. Figure 3.4 shows the amplitude of the AC magnetic field B0 measured with an

increasing peak-to-peak voltage amplitude at different frequencies from a function

65

generator. Due to the inductive impedance of the circuit√

R2 + ω2L2, the current

flowing through the Helmholtz coil and hence the magnetic field decreases with

increasing frequency. Because 1/f noise dominates over other noise sources at a

low frequency detection, a modulation frequency of 100 Hz was chosen to reduce

1/f noise and produce a reasonably strong magnetic field at the same time. This

choice was somewhat arbitrary, but we experimentally verified that the Faraday

rotation measured with different modulation frequencies were indeed the same

although it can affect the S/N ratio.

For a verification of our experimental procedure, we measured a 2 cm-long rod of

TGG at wavelengths of 633 nm, 850 nm, 1300 nm, and 1550 nm. As we discussed

in the previous section, the optical bias curve without applying magnetic field

was obtained by measuring the output from the balanced detector while rotating

the half-wave plate mounted on a computer-controlled step motor and fitting to

Eq. 3.12 as shown in Fig. 3.5.

From the optical bias curve obtained, the demodulated signal from the lock-in

amplifier representing the RMS value of amplitude modulation can be converted

to an angle of rotation for a given applied magnetic field. Note that the demodu-

lated signal for different magnetic flux density is averaged over one hundred data

points. As shown in Fig 3.6, we can determine the Verdet constant by fitting the

angles of rotation at different magnetic field values to a linear line, as indicated

in Eqs. 3.13-3.15. The measured Verdet constant of TGG is -141 rad/T · m at a

wavelength of 633 nm, which agrees well with a published value [25]. Note that

66

−25 −20 −15 −10 −5 0 5 10 15 20 25−0.2

−0.1

0

0.1

0.2

Angle (deg)

Pow

er (

V)

measuredfitted

Figure 3.5: Optical bias curve without a magnetic field applied for a 2 cm-long rod

of TGG sample obtained at wavelength 633 nm. The solid line is a fit to Eq. 3.12.

the Verdet constant of TGG is negative, indicating that this is a paramagnetic

material. From the measurement at different wavelengths as shown Fig. 3.7, we

can see that there is a wavelength dependence and the Verdet constant quickly

drops as wavelength increases. The measured values of V are -71.1, -21.2, and

-8.7 rad/T · m at wavelengths of 850 nm, 1300 nm, and 1550 nm, respectively.

Most recently, a very high Verdet constant from an organic polymer thin film

was reported and drew attention for potential applications to integrated optic de-

vices [31]. For this reason, thiophene based-organic samples identified as SJZ-87A

(25 µm), TK1V1292A320-A (100µm), and TK1V1292A320-N (100µm) were pro-

67

0 20 40 60 80 1000

0.01

0.02

0.03

Magnetic Field (Gauss)

Rot

atio

n (r

ad)

measuredfitted

Figure 3.6: Faraday rotation measured from a 2 cm long rod of TGG at a wave-

length of 633 nm by varying magnetic field flux densities and a fit to linear line.

0 20 40 60 80 1000

0.002

0.004

0.006

0.008

0.01

Magnetic Field (Gauss)

Rot

atio

n (r

ad)

633 nm 850 nm

1300 nm

1550 nm

Figure 3.7: Faraday rotation measured for a 2 cm-long rod of TGG at different

wavelengths

68

Figure 3.8: Chemical structure of SJZ-87A

vided by Prof. Seth Marder’s group at Georgia Tech. All three of them were drop-

dispensed and sandwiched by two glass substrates with a spacer. Figure 3.9 shows

the UV-Vis-IR transmission spectra of the samples measured with a Cary 5000

spectrophotometer.

We measured the Verdet constant of SJZ-87A at wavelengths of 1300 nm and

1550 nm, and the Verdet constants of TK1V1292A320-A and TK1V1292A320-N at

850 nm. Note that the substrate has a small but finite Verdet constant that must

be isolated. Hence, by measuring an angle of rotation from the substrate only,

the net rotation from the organic samples were determined. The measured Verdet

constant of the substrate was 2.9, 1.3, and 0.9 rad/T · m at 850 nm, 1300 nm, and

1550 nm, respectively. The values of V for SJZ-87A were 10.4 and 4.2 rad/T · m at

1300 nm and 1550 nm, respectively. For TK1V1292A320-A and TK1V1292A320-N,

the values were -4.7 and -4.2 rad/T · m at 850 nm, respectively.

The measured Verdet constants of these organic samples are about a half or

less than that of TGG, but new chemical structures based on polythiopene are

under development. To date, constructing integrated-optic isolators or circulators

69

200 400 600 800 1000 1200 1400 1600 1800 20000

10

20

30

40

50

60

70

80

90

100

TK1V1292A320_A

TK1V1292A320_N

Tra

nsm

issio

n (

%)

Wavelength (nm)

SJZ-87A

Figure 3.9: UV-Vis-IR transmission spectra measured with a Cary 5000 spec-

trophotometer for organic magneto-optic samples provided by Georgia Tech.

70

0 20 40 60 80 100 1200

0.5

1

1.5

2

2.5

3

3.5

4x 10

−5

Magnetic Field (Gauss)

Rot

atio

n (r

ad)

substrate at 1300nm

substrate at 1550nm

SJZ−87A at 1300nm

SJZ−87A at 1550nm

Figure 3.10: Faraday rotation measured for the substrates only and for the sample

labelled SJZ-87A at wavelengths of 1300 nm and 1550 nm.

0 20 40 60 80 100 1200

1

2

3

4

5

6

7

8x 10

−5

Magnetic Field (Gauss)

Rot

atio

n (r

ad)

substrate measuredsubstrate fittedTK1V1292A320−A measuredTK1V1292A320−A fittedTK1V1292A320−N measuredTK1V1292A320−N fitted

Figure 3.11: Faraday rotation measured for the substrates only and for the samples

labelled TK1V1292A320-A and TK1V1292A320-N at a wavelength of 850 nm.

71

from crystalline magneto-optic materials have been facing challenges. However,

organic materials can be potentially very useful because they can be fabricated

into waveguide type devices much easier than crystals or glass materials from

simple fabrication steps of spin-casting and photolithography.

3.4 Summary

In this chapter, we examined organic magneto-optic materials, which can be viable

alternatives to the chiral materials. In addition to polarization rotator, they also

can potentially be used for integrated-optic isolators or circulators using their non-

reciprocal polarization rotation. The Faraday effect from magneto-optic material

is another class of optical activity, and it is characterized by Verdet constant.

Measuring the Verdet constants of thin (10−100 µm) organic samples in a moderate

magnetic field (< 100 Gauss) can be very challenging. For this reason, we discussed

balanced homodyne detection technique that provides a highly sensitive technique

to measure an angle of polarization rotation as small as 0.5×10−6 rad. We validated

this experiment by measuring the Verdet constant of a 2 cm long rod of TGG and

confirming the published value. TGG is one of the most widely used material for a

free space optical isolator and is believed to have the highest Verdet constant until

orders of magnitude higher Verdet constant was observed in organic materials.

We demonstrated the Verdet constants of 10.4 and 4.2 rad/T · m at 1300 nm and

1550 nm, respectively, from an organic sample provided by Georgia Tech, which is

comparable to that of terbium gallium garnet. This unique observation can lead to

72

a new opportunity to build integrated-optic isolators or circulators from a simple

fabrication methodology including spin-casting and photolithography. However, in

order to build practical devices, much work needs to be done to develop materials

with even larger Verdet constant and detailed theory on the eigenmode properties

in magneto-optic waveguides.

73

Chapter 4

Optical Waveguide and Microring

Resonator from PFCB

4.1 Perfluorocyclobutyl (PFCB) copolymer

Among the many different kinds of organic polymer materials for passive photonic

waveguides, fluorinated polymers are of particular interest due to their signifi-

cantly low absorption loss at visible and infrared wavelengths. There are several

commercially available fluoropolymers including TeflonTM AF (tetrafluoroethylene

and perfluorovinyl ether copolymer, DuPont) [38], PolyguideTM (fluoroacrylate,

DuPont) [39], UltradelTM (fluorinated polyimide, Amoco) [40], PyralinTM (fluo-

rinated polyimide, HD Microsystems), CytopTM (perfluorovinyl ether copolymer,

Asahi Glass) [41], and TOPSTM(perfluorocyclobutyl, originally from DOW and

new PFCB-containing polymers under development at Tetramer Technologies,

L.L.C) [42]. However, these fluorinated polymers, in general, suffer from limita-

tions on processability including poor adhesion to other layers, limited solubility,

poor mechanical properties such as a large coefficient of thermal expansion (CTE)

and a large elongation constant that makes cleaving difficult.

The chemistry of PFCB polymers developed at Tetramer Technologies are

74

based on the thermal cyclopolymerization of bi- or tri-functional aryl trifluorovinyl

ether monomers as shown in Fig. 4.1. These PFCB polymers and their copolymers

are well suited for applications in the areas of integrated optics and photonics due

to the tailorability of their refractive indices (1.443–1.527 at 1550 nm) as well as

their high thermal, mechanical and optical stability, and high solubility (50 – 90

wt.%). In general, a material’s intrinsic contribution to the propagation loss at

telecommunication wavelengths is dominated by multi-phonon absorption due to

the excitation of molecular vibrations and Rayleigh scattering due to the local

density or composition fluctuations in the material. The absorption of light from

excited molecular vibrations in organic materials can be reduced significantly by

replacing C-H bonds with C-F bonds. A larger reduced mass of C-F bond results

in a lower vibrational resonant frequency (a longer resonant wavelength). There-

fore, the strength of absorption decreases by approximately an order of magnitude

between successive vibrational harmonics (sometimes referred to as vibrational

overtones) [43, 44]. As a consequence, the absorption loss in the fluorinated poly-

mer is greatly reduced at all the key communication wavelengths.

4.2 Low loss optical waveguide from PFCB copolymer

A low loss optical waveguide is one the most important building blocks in integrated-

optic communication systems because most passive and active devices consist of

simple straight and curved waveguide segments. In this section, we describe a de-

sign, fabrication and measurement of optical propagation loss of the PFCB-based

75

Figure 4.1: Thermal (a)polymerization and (b)copolymerization to PFCB-based

polymer from trifluorovinylaryl ether monomers. Five selective aryl substituent

are shown at the bottom. Courtesy of Tetramer Technologies, L.L.C.

76

polymer single mode waveguides.

4.2.1 Eigenmodes in dielectric waveguides

A single mode waveguide, which supports only one eigenmode for each polarization

state, is preferable in most photonic communication components. To briefly discuss

the design of a single-mode waveguide, let’s first consider the time-independent

wave equation for the electric fields,

∇2E + ∇(

1

n2∇(n2) · E

)

+ k20n

2E = 0 (4.1)

which can be obtained from the Maxwell’s equations 2.3, 2.4, 2.11, 2.12 with the

constitutive relations for non-magnetic dielectric materials given as

D = ǫE, (4.2)

B = µH, (4.3)

ǫ(x, y) = ǫ0n2(x, y), (4.4)

µ(x, y) = µ0. (4.5)

Because we consider the waveguide structures to have a piecewise constant re-

fractive index profile from a homogeneous dielectric medium rather than a graded

index profile and we assume an electric field with a longitudinal z-dependence of

e−jβz, the eigenvalue equation for the transverse components of the electric fields

can be written as

∇2tet + (k2

0n2 − β2)et = 0. (4.6)

77

Note that once the transverse components of the electric field, ex and ey, are

obtained by solving the above equation with proper boundary conditions, all the

other field components can be computed using Maxwell’s equations.

Except for a few special cases that can be solved analytically, such as a one-

dimensional planar waveguide extending infinitely in one direction or an optical

fiber with cylindrical symmetry, most dielectric optical waveguides require numer-

ical computation to solve the waveguide eigenvalue equations, such as the finite

element method (FEM) [45,46], the finite difference method (FDM) [47,48], and the

beam propagation method (BPM) [49–51]. Although there are several commercial

software packages available including Apollo Photonic Solutions Suite (APSSTM)

from Apollo Photonics, OptiBPMTM from Optiwave Corp., and BeamPROPTM

from RSoft Design Group, Inc., we performed a full-vectorial waveguide modal

analysis using a finite difference mode solver written in MATLAB code [52].

Two different types of guiding structures, the pedestal structure and the buried

channel structure, were considered with a PFCB core (n=1.498, with aryl sub-

stituent 2 in Fig. 4.1) and a PFCB cladding (n=1.458, with aryl substituent 3

in Fig. 4.1), both from Tetramer Technologies. Figure 4.2 shows the calculated

mode profile of the transverse electric field component |ex| and effective index

neff(= β/k0) for the fundamental TE mode for both structures. For the pedestal

waveguide, the dimensions of the core are 3.8 µm × 3.6 µm and the waveguide is

etched down 3.4 µm into the cladding layer. The calculated neff is 1.47571 and

1.47647 for the TE and TM, respectively. For the buried channel waveguide, the

78

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PFCB clad n=1.458

air n=1.0

3.8mm

3.6mm

TE mode |Ex| , neff = 1.4775

3.4mm

PFCB core n=1.498

(a)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PFCB clad n=1.458

PFCB core n=1.498

3.8mm

3.6mm

TE mode |Ex| , neff = 1.4835

(b)

Figure 4.2: Mode profile of the transverse electric field component |ex| for the

fundamental TE mode for two different types of straight waveguides, pedestal and

buried channel, using a full-vectorial finite difference mode solver. The orthogo-

nal field component |ey| is orders of magnitude smaller than |ex|, and hence it is

omitted. (a) Pedestal structure, (b) Buried channel structure.

79

dimensions of the core are 3.8 µm× 3.6 µm and neff is 1.48300 and 1.48297 for TE

and TM, respectively. Note that, in the buried channel waveguide, TE and TM

modes are nearly degenerate (nTEeff ≈ nTM

eff ) because it has a geometric configuration

close to that of a square and a symmetric index profile.

It is worth pointing out here that although the eigenmodes of an optical dielec-

tric waveguide are usually not pure TE or TM1, except in those two special cases

of a one-dimensional slab waveguide and an optical fiber that we mentioned above,

they are typically referred to as TE or TM modes since one of the two transverse

field components is orders of magnitude larger than the other. For example, in Fig.

4.2 the orthogonal field component |ey| is orders of magnitude (∼ 104) smaller than

|ex| in both cases and therefore it is not shown.

4.2.2 Fabrication of PFCB waveguides

This section describes the outline of the fabrication process of waveguides from

PFCB. Although the fabrication technique used here is not much different from

the ones used to build other integrated-optic waveguide devices because we adopted

an already existing conventional photolithography from the semiconductor IC in-

dustry, there are a few unique challenges to fabricating the waveguides from PFCB:

poor adhesion, a mask cracking problem, and deep etching required.

Figure 4.3 depicts an outline of the process steps to fabricate the PFCB waveg-

1In some articles or books, they are often referred to as HE/EH or Ex/Ey indicating they are

Hybrid modes. Note that the transverse electric field components |ex| and |ey| in Eq. 4.6 can

not be decoupled. Therefore, strictly speaking, TE/TM are not appropriate any more since the

solutions of eigenmode equations have both non-zero |ez| and |hz| [14, 53]

80

uides. We used Si substrates for a mechanical platform for the waveguides. In

general, fluorinated polymers have poor adhesion to other layers, which is a well-

known problem [54,55]. In order to improve the adhesion to the Si substrate, we de-

posited a 100 nm thick SiO2 layer from a Trion PECVD, after which we spin-casted

bisbenzocyclobutene (BCB, CycloteneTM 3022-35 purchased from Dow Chemical)

followed by soft curing at 210 C for 40 min as suggested by the vendor [56]. This

1 µm thick layer of BCB is believed to not only improve the adhesion strength to

SiO2/Si, but also relax the stress from the CTE mismatch between the PFCB and

the substrate. These two layers were omitted in Fig. 4.3 for clarity.

Next, solutions of PFCB cladding in mesitylene were spin-coated at a spin

speed to achieve a 7 µm thick layer and cured at 220 in the vacuum oven with

N2 environment. It is very difficult to put the PFCB core layer on top of the

PFCB cladding once the cladding layer is fully cured. This problem can be solved

with a partial curing of the cladding layer. However, when it is not cured enough,

the cladding is attacked by the mesitylene solvent contained in the core solution.

Therefore, adjusting the degree of partial curing to provide both good adhesion

and resistance to the solvent at the same time is very critical and this can be

found from trial-and-error. We have also found that dispensing mesitylene followed

by spin-drying and keeping the sample in a N2 environment for about one hour

before casting the PFCB core layer can provide an improved adhesion. After about

the 3.6 µm PFCB core layer was prepared, the sample was ready for etching the

waveguide.

81

1. Spin coat PFCB clad and cure

2. Spin coat PFCB core and cure

PFCB clad

PFCB core

Si

SiO2 : 3000 A

PR : 0.8 µm

3. Deposit SiO2 (RF Sputter)

4. Spin coat photoresist

5. Expose and develop PR

6. Pattern transfer to SiO2 by RIE

7. Pattern waveguides with RIE

8. Remove residual SiO2

3.6 µm

7.0 µm

7.0 µm

7. Pattern waveguides with RIE

8. Remove residual SiO29. Spin coat PFCB clad and cure

Pedestal

Buried channel

Figure 4.3: Outline of process steps to fabricate the PFCB waveguides. SiO2 and

BCB layers inserted for adhesion are not included for clarity.

82

Because a deep etch (∼ 7 µm in the case of the pedestal structure) is required

for a waveguide structure, a hard mask was used rather than a photoresist soft

mask. For the etching mask, about a 300 nm thick SiO2 layer was deposited on

PFCB dual layers using an RF sputter deposition in an argon plasma. We used an

AJA International magnetron sputter with a chamber pressure of 3mT and an RF

power of 250 W for 70 min, resulting in a deposition rate about 43 A/min. Initially,

a layer of SiO2 from PECVD was tried for the sacrificial hard etching mask, but

it caused a severe mask cracking problem. It is believed that the SiO2 from a

magnetron sputter has a lower packing density than the SiO2 from a PECVD,

hence it is less susceptible to cracking, which is likely due to the CTE mismatch.

We also found an alternative solution of using a metal bilayer of Ni/Au that can be

patterned by deposition and lift-off. The Ni/Au etching mask can be particularly

useful for an etching mask in the case that the SiO2:cladding selectivity is poor,

for example, when a thermal SiO2 layer is used for a lower cladding.

For waveguide patterning, an 800 nm thick layer of OIR 906-10 photoresist was

exposed with a 5× GCA i-line projection aligner and developed in OPD 4262 for

60 seconds. The optimized exposure condition was found to be an expose duration

of 0.13 s and a focus offset of 0 through a focus/exposure test. Next, the waveguide

pattern was transferred to the sputtered SiO2 by reactive ion etching using a

Plasmatherm RIE system in a CHF3/O2 plasma with a flow rate of 18 sccm of

CHF3 and 2 sccm of O2, a chamber pressure of 40 mT and an RF power of 200 W,

which resulted in an etch rate of 40-45 nm/min. It should be noted that a slight

83

over etch into the core layer is required to ensure that the SiO2 has completely

cleared. After patterning the SiO2 mask, a waveguide was etched by RIE in an

Ar/O2 plasma with a flow rate of 20 sccm of Ar and 10 sccm of O2, a chamber

pressure of 20 mT, and an RF power of 250 W, which resulted in a PFCB etch

rate of 150-160 nm/min. The PFCB:SiO2 selectivity greater than 25:1. If desired,

any residual photoresist on top of SiO2 could be removed easily in acetone, but

it is not required. After the PFCB etching was completed, the remaining SiO2

was removed in a buffered oxide etch (BOE). For the buried channel waveguide

structure, the PFCB clad layer was spun and cured to serve as a top cladding layer

after patterning the core. For the final step, devices were cleaved to form smooth

end facets allowing end-fire coupling for measurements.

Figure 4.4 shows the scanning electron micrograph of the PFCB pedestal waveg-

uide. As shown in the figure, the PFCB core and clad layer are distinguishable by

exhibiting composition-dependent side wall roughness.

4.2.3 Loss measurement of straight waveguides

The loss measurement of a straight waveguide is basic but critical because one often

wants to isolate the propagation loss from total waveguide insertion loss. This

can be very helpful in designing more advanced integrated-optic devices. From

several available methods for waveguide-loss measurement, we chose the cut-back

method [57], which is the most simple and straightforward technique. A drawback

of this method is that it is a destructive measurement method requiring successive

84

Figure 4.4: Scanning electron micrograph of the PFCB pedestal waveguide.

cleaving [58]. Another popular way to measure the loss is using the properties

of Fabry-Perot resonance, because the two end facets of the waveguide form a

resonant cavity [59, 60]. However, for polymer waveguides, the reflectivity at the

air/waveguide interface is rather small and the cleaved surfaces are not quite as

perfect as those of semiconductor waveguides, so the Fabry-Perot method is less

reliable.

Figure 4.5 shows the measured total insertion loss at a wavelength of 1550 nm

for the TE mode at four different waveguide lengths achieved by sequential cleav-

ing. A single mode fiber with a conical tip (with a focal length of ∼ 8 µm) was

used to allow efficient end-fire coupling between the optical fiber and the waveg-

uide. Once the total transmission loss was measured, the coupling loss and the

propagation loss can be separated using a linear regression – the coupling loss is

85

0.0 0.2 0.4 0.6 0.8 1.0 1.22.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

pedestal

(b) 0.3 x + 2.82

(a) 1.1 x + 4.63

In

sert

ion L

oss (

dB

)

Length (cm)

buried channel

Figure 4.5: Loss measurement of PFCB straight waveguides using cutback method

for TE mode. The measured propagation loss is 1.1 dB/cm and 0.3 dB/cm, and

coupling loss is 4.63 dB and 2.82 dB for (a) pedestal and (b) buried channel waveg-

uide, respectively.

from the insertion loss at the zero waveguide length, and the propagation loss is

from the slope. The measured propagation loss was 1.1 dB/cm and 0.3 dB/cm, and

coupling loss was 4.63 dB and 2.82 dB for the pedestal and buried channel waveg-

uides, respectively. The difference in propagation loss between the two waveguides

comes mostly from a difference in scattering loss: In general, scattering loss is

larger at the waveguide side walls than at the horizontal (upper and lower) in-

terfaces because the presence of surface roughness is dominant at the side walls

86

from reactive ion etching (RIE) procedures [61, 62]. For the same reason, the side

wall roughness exhibits the polarization dependent loss (PDL) – the TE loss is

typically slightly higher than the TM loss. Furthermore, the surface-roughness-

induced scattering loss is particularly high in the case of tighter confinement of the

mode in the core as opposed to the case of weak confinement because of different

index contrast [39]. On the other hand, the coupling loss is smaller in the buried

channel waveguide because of the difference in mode mismatch between the fiber

and the waveguide [63].

4.3 PFCB Microring resonators

Devices based on a microring resonators have been actively investigated by many

research groups due to their simple configuration and yet highly interesting multi-

ple functionality. Some applications include phase shifters, optical channel drop-

ping filters (OCDF), optical add-drop (de)multiplexers, switches, photonic logic

gates, amplifiers and lasers. There are a number of excellent journal articles and

theses describing the comprehensive historical background and fundamental the-

ory on microring and microdisc resonators [64–68]. Therefore, we begin by de-

scribing a very succinct summary of the basic theoretical principles of the ring

resonator rather than a comprehensive summary, and then we discuss the design

considerations and characterization of microring resonators fabricated from PFCB

copolymers.

87

κ, τEi Et

Erb Era

(a)

κ1 , τ1Ei Et

Erd Era

κ2 , τ2

Ed Ea

Erc Erb

Drop

Input

Add

Through

(b)

Figure 4.6: Schematic diagram of microring resonators. (a) all-pass and (b) add-

drop configuration.

4.3.1 Basic principles

The microring resonator is, by nature, very similar to a Fabry-Perot resonator

in many aspects. The most simple configuration of the microring device is such

that the ring is coupled to one or two adjacent waveguides through ‘evanescent

field coupling’ as shown in Fig. 4.6. The former is often referred to as an “all-pass

88

filter” and the latter is commonly called an “add-drop filter”.

By adopting the scattering matrix formulism based on energy conservation and

time-reversal symmetry [69], the relationships between the optical electric fields for

the add-drop filter, shown in Fig. 4.6, can be written as

Ea

Et

=

τ1 −jκ1

−jκ1 τ1

Erd

Ei

, (4.7)

Ed

Erc

=

τ2 −jκ2

−jκ2 τ2

Erd

Ei

. (4.8)

We can also write equations describing the circulation condition along the ring,

Erb = Era ·√

A · ejφ/2, (4.9)

Erd = Erc ·√

A · ejφ/2, (4.10)

where A = exp (αRL/2) represents the field loss (or gain) after one round-trip,

φ = k0neffL is the round-trip phase change, αR is the power loss coefficient in the

ring, neff is the effective index of the waveguide, L = 2πR is the circumference of

the ring, κ(1,2) are the field coupling coefficients between the bus and the ring, and

τ(1,2) are the field transmission coefficients. We also assume a lossless coupling, in

which case τ(1,2) =√

1 − κ2(1,2).

Therefore, from Eqs. 4.7–4.10, the optical fields at drop port Ed and through

89

port Et can be expressed as,

Et =τ1 − τ2Aejφ

1 − τ1τ2AejφEi +

−κ1κ2

√Aejφ/2

1 − τ1τ2AejφEa, (4.11)

Ed =−κ1κ2

√Aejφ/2

1 − τ1τ2AejφEi +

τ2 − τ1Aejφ

1 − τ1τ2AejφEa. (4.12)

When there is no input signal at the add port, i.e. Ea = 0, Eqs. 4.11 and are

reduced to

Et

Ei

=τ1 − τ2Aejφ

1 − τ1τ2Aejφ, (4.13)

Ed

Ei

=−κ1κ2

√Aejφ/2

1 − τ1τ2Aejφ. (4.14)

The optical field at the through port Et for the all-pass configuration can be

obtained by putting τ1 = τ and κ1 = κ, and letting κ2 = 0 in Eq. 4.13 because there

is no second waveguide adjacent to the ring. Then, for the all-pass configuration,

we have

Et

Ei

=τ − Aejφ

1 − τAejφ(4.15)

=A − τejφ

1 − τAejφ· ej(π+φ). (4.16)

The phase of Et can be obtained from Eq. 4.16 and can be written as,

Φ =τ − Aejφ

1 − τAejφ. (4.17)

For the all-pass configuration, when the ring resonator is lossless, i.e. A = 1, (but,

in practice, this has to be the negligible loss case, A ≈ 1) the output power is unity

(100% transmission) at all wavelengths regardless of the coupling coefficient κ. On

90

1535 1540 1545 1550 1555 156010

2

10 1

100

Wavelength (nm)

No

rma

ilize

d in

ten

sity

, |

Et|2

Figure 4.7: Calculation of a typical spectral response of all-pass filter with neff =

1.477, R = 30 µm, and τ = 0.9 for when A = 1 and A = 0.9 representing the

lossless case and the round-trip power loss of 19%, respectively.

the other hand, when the ring has a finite amount of loss that satisfies “critical

coupling”, in which the round trip field loss of the ring equals the field coupled into

the ring from the bus waveguide, i.e. A = τ , the output power has high extinction

at the resonant wavelengths. This could be used as an optical channel-drop filter

(OCDF), also referred to as a notch filter. Figure 4.7 shows the calculation of a

typical spectral response of an all-pass filter with neff = 1.477, R = 30 µm, and

τ = 0.9 when A = 1 and A = 0.9 representing the lossless case and the round-trip

power loss of 19%, respectively.

Although the all-pass configuration with a negligible loss is not useful as a

91

φ = δ (ωT)

Pha

se S

hift,

θ

2π

π

0 0 −π/4 π/4

τ = 0.99

τ = 0.90

τ = 0.95

A = 0.99

Figure 4.8: Calculated phase response near a resonance in all-pass configuration

for three different τ ’s when the loss is very small, A = 0.99.

channel drop filter, its interesting phase response near the resonant wavelength

(φ ≈ 0), as shown in Fig. 4.8, makes it possible for use as a π phase shifter. This

can be utilized as a notch filter by incorporating it into one arm of a Mach-Zehnder

interferometer [66]. The add-drop configuration is mathematically equivalent to

a Fabry-Perot interferometer with two input and two output ports. Using the

narrow-band amplitude filter response from its high wavelength selectivity, one

can drop a particular frequency band from the input port via the drop port and

add another band from the add port to the through port. Figure 4.9 shows a

typical spectral response of an add-drop filter with neff = 1.477, R = 30 µm,

A = 0.95, and τ1 = Aτ2 = 0.9 satisfying the critical condition. Resonators are

92

1542 1544 1546 1548 1550 1552 1554 1556 1558 156010

3

10 2

10 1

100

Wavelength (nm)

Norm

aili

zed in

tensi

ty

Figure 4.9: Calculation of a typical spectral response of add-drop filter with neff =

1.477, R = 30 µm, A = 0.95, and τ1 = Aτ2 = 0.9 satisfying the critical condition.

typically characterized by their free spectral range ∆νFSR (or ∆λFSR), resonance

width ∆νFWHM (or ∆λFWHM, the full width at half maximum), quality factor Q,

and finesse F . The free spectral range is the separation between two successive

resonances expressed as,

∆νFSR =c

Lneff

=1

Tin frequency, (4.18)

∆λFSR =λ2

0

Lneff

in wavelength, (4.19)

and the finesse F is defined as

F =∆νFSR

∆νFWHM

(4.20)

93

The quality factor Q, describing the rate at which the resonator dissipates its stored

energy and the sharpness of a resonance, is defined as the resonance frequency

divided by the bandwidth, Q = ν0/∆νFWHM, and is related to the finesse by

Q = LneffF/λ.

4.3.2 Design considerations

Mode analysis

In section 4.2.1, we discussed the eigenmodes of the dielectric waveguides and nu-

merically computed the eigenvalue equation to find the eigenmodes for a given

waveguide geometry. The numerical mode profile that we obtained can also be

used to compute the scattering loss at the waveguide boundary or the coupling

efficiency between two adjacent waveguides. To construct the microring resonator

devices, the pedestal waveguide structure was adopted simply because it allows

the most tightly confined guiding geometry.

Bending Loss

The most commonly used practical figure of merit for a ring resonator is the quality

factor or finesse, which is directly related to the radius of the ring, the propaga-

tion loss in the ring, and the coupling efficiency between the ring and waveguide.

Obtaining loss as low as possible is essential because the loss directly dictates the

resonance bandwidth (FWHM), hence Q and F of the ring resonator. Two main

contributions to loss in the ring are bending loss and scattering loss. For a given

index contrast of the waveguide, one can simply increase size of the ring and in-

94

crease etching depth of the pedestal structure in order to minimize the bending

loss.

For the PFCB microring devices, the waveguide dimensions in Fig. 4.2(b) and a

radius of curvature R in the range of 25–40µm are chosen from detailed numerical

modelling for the bending loss. There is a simple model to calculate bending loss

in planar slab waveguides based on Fraunhoffer diffraction theory [70]. This model

can be very useful especially in the early phase of the design by providing a good

starting point for determining a radius of curvature and width of the waveguide.

However, for most practical waveguides with a 3D structure confining the lightwave

both horizontally and vertically, this is only an good approximation.

We performed a 3D full vectorial FD calculation incorporating a conformal

transformation and a complex perfectly matched layer (PML) [71–73]. Figure 4.10(a)

shows that a radius R of 30 µm with an etch depth 7 µm is estimated to yield neg-

ligible bending loss (< 10−3 dB). Figure 4.10(b) shows the etch depth dependence

from the calculation of the bending loss with several different etch depths when

R = 30 µm. The round-trip power loss A2(= exp (αRL)) as a function of etch

depth h can be fitted to an exponential curve, i.e. A2 = 1 − e−2.2h. It should

be noted that this dependence could almost be predicted from the fact that the

field decay constant in the lower cladding layer, k0

√

n2eff − n2

s, is about 1.13 when

neff = 1.4775, and the power decay constant is twice as much.

95

µm

µm

A2 = 0.99978

-60dB

PML layer

(a) Bending loss when R = 30µm

4 4.5 5 5.5 6 6.5 70

0.2

0.4

0.6

0.8

1

Etch depth ( µm)

Ro

un

d t

rip

po

we

r lo

ss (

A2)

e-2.2h

(b) Etch depth dependence

Figure 4.10: 3D full vectorial calculation for a bending loss. (a)negligible bending

loss < 10−3 dB when a radius R of 30 µm with a etch depth of 7 µm, and (b)Etch

depth dependence of the bending loss when R = 30 µm. The round-trip power

loss A2 as a function of the etch depth h can be fitted to an exponential curve,

A2 = 1 − e−2.2h.

96

Coupling efficiency

One of the most important design parameters for constructing a practical microring

resonator device is the coupling efficiency – the amount of power transferred from

one waveguide to another when they are in close proximity – between the waveguide

and the ring. From this, we can determine the physical size of the gap required

for the critical coupling condition.

In a system consisting of two adjacent waveguides, the coupling constant be-

tween the two can be numerically computed from the mode profile we obtained

before and the coupled-mode theory [52, 74]. A very comprehensive and excellent

summary on non-orthogonal coupled-mode theory can be found in [52]. Therefore,

in this section, we present only the computation results for the coupling constant

µ in the coupled-mode equation,

d

dz

a1(z)

a2(z)

= −j

β1 µ12

µ21 β2

a1(z)

a2(z)

, (4.21)

where β1 and β2 are the propagation constants of waveguide 1 and waveguide 2,

respectively.

When the two waveguides are identical to each other, we can calculate the

coupling constant in a more direct way by considering the coupled waveguide

system as a whole rather than as two separate waveguides. The approximate

normal modes of the system can be expressed as the symmetric and antisymmetric

97

g

PFCB clad, n = 1.458

PFCB core, n = 1.498

air, n = 13.6 µm

3.6 µm

3.4 µm

Figure 4.11: Cross-sectional mode profile of two parallel waveguides separated by

an edge-to-edge distance g.

combinations of the isolated waveguide modes as

βs = β + µ, vs =1√2

=

+1

+1

, (4.22)

βa = β − µ, va =1√2

=

+1

−1

. (4.23)

Therefore, the coupling constant µ can be written as

µ =1

2(βs − βa). (4.24)

Figure 4.11 shows the mode profile of two parallel PFCB pedestal waveguides

separated by an edge-to-edge distance g, which was used to compute the coupling

98

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 210

−6

10−5

10−4

10−3

10−2

Coupling gap (µm, edge to edge)

Cou

plin

g co

nsta

nt,

µ (

µm

−1 )

Non−orthogonal Coupled Mode TheoryExact Mode Analysis

Figure 4.12: Calculation of coupling constant between two PFCB pedestal waveg-

uides as a function of waveguide separation g.

constant. Figure 4.12 shows the calculated coupling constant µ between two PFCB

pedestal waveguides as a function of waveguide separation g from both the non-

orthogonal coupled theory and the direct mode analysis. Note that the coupling

constant µ drops exponentially with increasing separation g, as we expect from

the fact that electric field decays exponentially in the cladding region. Also note

that the direct modal evaluation agrees very well with the coupled mode theory,

especially when g > 0.3 µm.

In order to find the field coupling coefficient κ in Eqs. 4.13–4.16, we consider

99

the transfer matrix, which is the solution of the differential coupled mode equation:

a1(z)

a2(z)

= ejβ0z

cos σ(z) − sin σ(z)

− sin σ(z) cos σ(z)

a1(0)

a2(0)

, (4.25)

with

σ(z) ≡∫ z

0

µ(z) dz, (4.26)

where σ(z) represents the integrated coupling when the two waveguides approach

and separate gradually as in the microring resonator. If light is launched into

waveguide 1 at z = 0, then the relative power in the two waveguides as a function

of z is given by,

∣

∣

∣

∣

a2(z)

a1(0)

∣

∣

∣

∣

= sin σ(z), (4.27)

∣

∣

∣

∣

a1(z)

a1(0)

∣

∣

∣

∣

= cos σ(z). (4.28)

Therefore, the coupling coefficient κ between the ring and the bus waveguide is

κ = sin σ(z) (4.29)

≈ σ(z) when σ(z) ≪ 1. (4.30)

4.3.3 Experimental characterization

The fabrication process to build the PFCB microring resonators is almost identical

to the one described in section 4.2.2. Effort was made to achieve a coupling gap

g as small as possible. The limitation on the coupling gap comes mainly from

the following two causes. First, the smallest feature size is fundamentally limited

by the wavelength of the light source employed in the i-line (λ = 365 nm from

100

Mercury lamp) projection aligner we used for photolithography. Second, the tight

confinement of guided light to reduce the bending loss requires a deep etch into

the cladding layer as we discussed earlier. In general, a high RF power and low

chamber pressure can increase the anisotropicity (or verticality) in RIE etching,

but the side wall etching to a certain degree is unavoidable. Figure 4.13 shows

a scanning electron micrograph of a PFCB add-drop microring resonator with a

radius of 25µm and a coupling gap of 0.45 µm. A coupling gap down to 0.4 µm

with our stepper is possible, but it is very difficult to achieve.

Figure 4.14 illustrates the experimental setup for characterizing the microring

devices. We used a Photonetics tunable laser and/or an erbium doped fiber am-

plifier (EDFA) for light sources. Light was coupled into the waveguide using a

conical-shaped lensed fiber, which has a focal length of approximately 8 µm. A

signal emerging from either the through port or the drop port is collected by a

lensed fiber and directed to a detector and a spectrum analyzer. Both the input

and output fibers are mounted on 3-axis translation stages from Newport, allowing

accurate and stable alignment.

Using a broad band amplified spontaneous emission (ASE) from an EDFA as

a light source can be very useful for device alignment because a spectrum ana-

lyzer allows us to monitor the spectral response of the device in nearly real-time.

Figure 4.15 shows the dropped signal at the drop port of an add-drop filter with a

radius of 25µm when the ASE light is coupled into the input port. As shown in the

figure, the measured free spectral range λFSR = 9.3 nm. Note that the unpolarized

101

µ

µ

Figure 4.13: Scanning electron micrograph of PFCB ring resonator, with a radius

R = 25 µm, and a coupling gap g = 0.45 µm.

102

Tunable Laser Isolator

Polarization Controller

Lensed Fiber

Power-meterDetector

EDFA

Wavemeter

Lensed Fiber

Spectrum

Analyzer

GPIB

Figure 4.14: Experimental setup for microring characterization.

ASE source shows the resonances for both TE and TM mode, and the different

signal intensities for TE and TM can be explained by the polarization dependent

scattering loss (PDL) and the difference in coupling constants.

Figure 4.16 plots the normalized signal of the device at the through and drop

port at the resonance at 1531.05 nm from a tunable laser input. As shown in

the figure, the maximum extinction ratio at the throughput port is 4.87 dB, and

the calculated quality factor Q and finesse F are approximately 8544 and 55,

respectively.

In addition, we observed a resonance shift with increasing EDFA input power.

When the total ASE power was increased from 4 dBm to 12 dBm, a resonance shift

103

1520 1530 1540 15500.0

5.0x10-8

1.0x10-7

1.5x10-7

2.0x10-7

Wavelength (nm)

Inte

nsity

0.0

1.0x10-9

2.0x10-9

3.0x10-9

TE

TM

R = 25 µm

∆λFSR = 9.3 nm

Figure 4.15: Measured spectral response at the drop port of a add-drop filter with

a radius 25µm with ASE input (solid line). The measured free spectral range λFSR

is 9.3 nm for TE mode. EDFA spectrum (dotted line) is also plotted as a reference.

104

1530 1530.5 1531 1531.5 15320

0.2

0.4

0.6

0.8

1

Wavelength (nm)

Norm

aliz

ed in

tensi

ty

Throughput

Drop

Figure 4.16: Normalized signal of the device at the throughput and drop port at

the resonance at 1531.05 nm from a tunable laser input.

of 25 pm was obtained as shown in Fig. 4.17(a). We speculate that the origin of

this resonance shift is thermal, based on the observation of the negative thermo-

optic coefficients, dn/dT , of the PFCB polymers as shown in Fig. 4.17(b). The

thermo-optic coefficient was obtained by measuring refractive indices at different

temperatures with a MetriconTM that was custom fitted with temperature con-

trol. This negative dn/dT that lowers the mode effective index of the waveguide

can qualitatively explain the blue-shift of the resonance with increasing EDFA

input power. Similar effects were observed in microrings fabricated from benzocy-

clobutene (BCB) polymer, and optical bistable switching based on the resonance

105

shift was demonstrated by other members in our group [75,76]. They are currently

investigating the subject for more complete understanding of the physical origin.

4.3.4 Ultrafast all-optical switching experiment

More than two decades ago, ultra-fast optical control of microwave and millimeter

wave propagation was reported [77, 78]. This ground work demonstrated phase

shifting, switching, and modulating in semiconductor waveguides using a picosec-

ond laser pulse. Likewise, the light signal out of a microring can also be modulated

or switched by changing the optical properties of the material with an optical pump

that either co-propagates through the ring together with a probe signal [79, 80],

or directly illuminates the ring from above. All-optical lightwave switching in

GaAs semiconductor microring devices was also achieved by injecting free carriers,

generated by absorbing photons with an energy greater than the band gap [81].

In this section, we show that all-optical switching in polymer microrings is

possible from preliminary experimental results. The fundamental physics of car-

rier dynamics in polymer devices is substantially different from the direct optical

excitation of electrons and holes for the GaAs semiconductor microring devices.

Although it is not clearly understood at this point, the all-optical switching we

observed can still be explained by carrier injection (excitons and free carriers),

as evidenced by an earlier report on the transient photoconductivity in a BCB

polymer film induced by a femtosecond laser pulse [82].

For the switching experiment, we used a microring-loaded Mach-Zehnder inter-

106

1533.2 1533.4 1533.6 1533.8 15340

0.5

1

1.5

2x 10

−8

Wavelength (nm)

Inte

nsity

4 dBm 6 dBm 8 dBm10 dBm12 dBm

(a) ASE input power vs. resonance shift

20 40 60 80 100 1201.45

1.46

1.47

1.48

1.49

1.50

1.51

PFCB clad, dn/dT = -5.95x10-5 oC

-1

PFCB core, dn/dT = -9.13x10-5 oC

-1

Index o

f R

efr

action

Temperature (oC)

(b) Thermo-optic coefficient, dn/dT

Figure 4.17: (a) Resonance shift with increasing ASE input power. Maximum

resonance shift 25 pm obtained. (b) Thermo-optic coefficient, dn/dT , measured

using a MetriconTM that was custom outfitted with temperature control.

107

ferometer (MR-MZI) – a microring coupled to one arm of an MZI – fabricated from

PFCB, which is schematically shown in Fig. 4.18(a). In the experiment, a train of

mode-locked Ti:Sapphire laser pulses with a pulse width of 100 fs, a repetition rate

of 1 KHz at a wavelength of 400 nm was used for the optical pump beam. When

these pump pulses directly illuminate the ring from above, it causes a temporal

blue shift of the microring resonance as schematically shown in Fig. 4.18(b). When

the probe beam is initially tuned to a wavelength slightly shorter than a ring res-

onance (λ1p < λR in Fig. 4.18(b)), the probe beam experiences a rapid decrease

in transmission temporally due to the resonance shift from λR to λ′R, and hence

exhibits High-Low-High modulation behavior. On the other hand, when the probe

beam is initially tuned to a resonance (λ2p = λR in Fig. 4.18(b)), the probe beam

experiences a rapid increase in transmission, and hence exhibits Low-High-Low

modulation behavior. As a result, the train of pulses modulates the probe signal

in time as shown in Fig. 4.18(d) and (e), respectively.

In addition to this amplitude modulation from the resonance shift, the opti-

cal pump pulse changes the phase near a ring resonance as we discussed in sec-

tion 4.3.1. Fig. 4.18(c) illustrates the phase response across the resonance. For

an MR-MZI, the temporal π phase shift around a resonance of the ring resonator

can destructively interfere with the light in the second arm, therefore, as in the

amplitude modulation, the train of pulses modulates the probe signal in time as

shown in Fig. 4.18(d) and (e), depending on whether the probe signal is initially

on-resonance or off-resonance. From the combined effect of amplitude modulation

108

Low-High-Low t

I

I

λλRλR'

H

φ = δ (ωT)

Ph

ase

Sh

ift,

θ

2π

0

0 /4 /4 −π π

t

I

LH L

H

H

(a)

(b) (c)

High-Low-High

(d) (e)

π

λp

1 λp2

L

Figure 4.18: Schematic picture of all-optical switching experiment. (a)An optical

pump with a 100 fs Ti:Sapphire laser pulse with an energy of 500 pJ/pulse at 400 nm

directly illuminates the ring from above. The resulting modulation behavior comes

from the combined effect of (b) amplitude and (c) phase modulation. (d),(e)

Therefore, the probe signal is modulated in time H-L-H or L-H-L depending on

the initial probe wavelength.

109

and phase modulation, the switching efficiency can be greatly increased around

the resonance [83,84]. However, it should be noted that this phase contribution is

significant only when the one arm of the MZI is over-coupled to the ring and the

loss in the ring is negligibly small.

The transfer response at the output of an MR-MZI is given by

Et =Ei

2

|Er|ejΦej∆φ + 1

(4.31)

where Er and Φ were given by Eqs. 4.16 and 4.17, respectively, and ∆φ is the phase

imbalance between the two arms of the MZI. Note that a finite imbalance in length

between two arms of an MZI resulting from fabrication imperfection is inevitable

and the phase difference (∆φ = k0neff∆d) from the length imbalance (∆d) can

greatly affect the spectral response of an MR-MZI over a wide range of wavelengths.

The spectral response of an MR-MZI shown in Fig. 4.19 is exaggerated for the case

of a length imbalance, ∆d = 5 µm, and a ring radius, R = 10 µm. As shown in

the figure, when the phase difference between the two arms is close to ∆φ = π,

the ring resonances exhibit ‘peaks’ rather than ‘dips’, and when 0 < ∆φ < π, the

resonance dips show asymmetric profiles. However, the ∆d can easily be controlled

to less than 100 nm, and therefore the response from an MZI without a microring

is essentially flat over a narrow range of wavelengths in most cases.

Figure 4.20 shows the preliminary experimental data showing the modulation

(or switching) response of the device (R = 50 µm) when a 100 fs Ti:Sapphire laser

pulse with an energy of 500 pJ/pulse was used for an optical pump beam. By

110

1300 1400 1500 1600 1700

0.2

0.4

0.6

0.8

Wavelength (nm)

Norm

aliz

ed in

tensi

ty

∆φ = 0

∆φ = π

Figure 4.19: Spectral response of an MR-MZI with a length imbalance between

two arms of the MZI, ∆d = 5 µm, which is exaggerated for clarity. The microring

is critically coupled and has a round-trip loss, A = 0.95, an effective index, neff =

1.477, and a radius of the ring, R = 10 µm.

111

sweeping the wavelength of a probe beam across a ring resonance (λR ≈ 1554.9 nm)

and detecting the time variation of the transmitted signal, we can estimate how

much the resonance is shifted. As shown in the figure, when the probe beam

was initially tuned to a wavelength slightly shorter than a ring resonance, and

progressively scanned towards the resonance, the dropped probe beam exhibited

High-Low-High modulation behavior. On the other hand, when the probe beam

was initially tuned to the resonance and scanned away from the resonance, the

signal is modulated as Low-High-Low. As discussed before, this can be explained

by the temporal change in the optical properties of the PFCB polymer. A response

pulse width of about 30 ps, which is primarily limited by the sampling scope used

in the measurement, and a maximum modulation depth of 3.8 dB was obtained.

Note that the response window is wider in the case that the probe wavelength was

scanned away from the ring resonance because of the asymmetric resonance profile

arising from length imbalance as shown in Fig. 4.19.

The estimated resonance shift ∆λ is around 0.6 nm, leading to a maximum

refractive index change, ∆n ≈ −5.7×10−4. From the pulse energy of 500 pJ/pulse

used in the experiment, if this resonance shift results from the intensity dependent

nonlinear refractive index, n = n0 +n2I, from 3rd order nonlinear optical effect, an

n2 on the order of −8× 10−18 m2/W, and a |Reχ(3)| on the order of 4× 10−9 esu

would be required. This large value of χ(3) is much higher than one might expect

from PFCB, therefore it can be speculated that the observed behavior results from

exciton and/or free carrier effects as we observed in the GaAs microring devices.

112

0 500 1000 1500 2000

0

2

4

6

8

10

12

Outp

ut S

ignal (m

V)

Time (ps)

1554.00 nm

1554.25 nm

1554.50 nm

1554.75 nm

(a) towards resonance

0 500 1000 1500 2000

0

2

4

6

8

10

12

Outp

ut S

ignal (m

V)

Time (ps)

1555.00 nm

1555.25 nm

1555.50 nm

1556.0 nm

(b) away from resonance

Figure 4.20: Measured switching response of the MR-MZI when the ring was

illuminated by a 100 fs Ti:Sapphire laser pulse with an energy of 500 pJ/pulse at

400 nm. The probe signal was swept through one of the ring resonances, λR ≈1554.9 nm.

113

However, the switching efficiency is about two orders of magnitude worse than

for the GaAs devices, mainly from the following two reasons. First, the carrier

generation efficiency is worse than the one from the conduction–valence interband

transition from linear absorption in the GaAs device. Second, a phase contribution

to switching is not significant because the ring in our case is under-coupled to the

MZI and the loss present in the ring cannot be neglected. Figure 4.21 shows a

phase response around a resonance for the all-pass configuration with a round-

trip field loss A = 0.9 when the field transmission coefficients are τ = 0.95 and

τ = 0.98. When the ring is lossy, particularly when the coupling efficiency is very

small compared to the loss (for example, solid line in the figure, a round-trip power

loss of 19% and a power coupling efficiency of 4%), the effective phase shift Φ shows

a very different behavior from the one we discussed in section 4.3.1 and it is much

smaller than a radian. Therefore, the phase modulation represented by ejΦ term

in Eq. 4.31 is not significant.

4.4 Summary

In this chapter, we investigated low-loss waveguides and microring resonators fabri-

cated from perfluorocyclobutyl (PFCB) copolymer. Fluorinated polymers are one

of the most attractive materials to construct photonic passive waveguides because

they have substantially low absorption loss at telecommunication wavelengths.

Low-loss waveguides are one of the most important building blocks in integrated-

optic communication systems because most passive and active devices consist of

114

−0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8

−0.2

−0.1

0

0.1

0.2

Normalized detuning, δ(ωT)

Effe

ctiv

e ph

ase

shift

, Φ

A = 0.9

τ = 0.98

τ = 0.95

Figure 4.21: Calculated phase response near a resonance in all-pass configuration

when the ring is lossy and under-coupled, a round-trip field loss A = 0.9, the field

transmission coefficients are τ = 0.95 and τ = 0.98.

115

simple straight and curved waveguide segments. For example, microring resonator

based devices have very simple configurations of straight waveguides and circular

rings, and the loss present in the ring is the most important parameter determin-

ing the device performance. For these reasons, this chapter was devoted to the

design, fabrication, and characterization of those devices using PFCB polymers.

We demonstrated straight waveguides with propagation losses of 0.3 dB/cm and

1.1 dB/cm for a buried channel and pedestal structures, respectively, and a mi-

croring add-drop filter with a maximum extinction ratio of 4.87 dB, quality factor

Q = 8554, and finesse F = 55. In addition, we showed that all-optical switching

with the PFCB microring resonator is possible when it is optically pumped by

a femtosecond laser pulse with sufficient energy. From a microring-loaded Mach-

Zehnder interferometer (MR-MZI), we demonstrated that a modulation window of

30 ps and modulation depth of 3.8 dB from an optical pump with a pulse duration

of 100 fs and pulse energy of 500 pJ when the signal wavelength is initially tuned

to one of the ring resonances.

116

Chapter 5

Organic Thin-Film Photodetector

based on Bulk Heterojunction

Ever since the discovery of organic semiconducting polymers1, new avenues for

their use as replacement for inorganic semiconductors have been explored. This

interest is fueled by their potential capabilities and possibilities for modern elec-

tronic and photonic devices, which include light emitting diodes (LEDs) [85, 86],

photovoltaic cells (PVs) [87–90], photodetectors (PDs) [91–94], and field effect

transistors (FETs) [95–98]. Among these, development of OLEDs has nearly ma-

tured and they were recently employed in the commercial flat panel displays for

small electronic appliances such as cell phones and PDAs. The commercial success

of this technology provided further spur to research on organic photovoltaic cells

in particular and organic photodetectors to a lesser extent because of their very

low manufacturing cost and substrate flexibility. Because solar energy conversion

in OPV cells and light detection in OPDs share a lot of common basic physics,

the work on organic thin film photodetectors that we discuss in this chapter has

benefited from the ideas developed in the study of OPV cells.

1This pioneering work on the discovery and development of electrically conductive polymers

by Alan J. Heeger, Alan G. MacDiarmid, and Hideki Shirakawa was awarded the Nobel Prize in

Chemistry in 2000 [3]

117

5.1 Basic principles

Charge carrier transport and light absorption in organic semiconducting materials

are associated with electronic bands formed by sp2-hybridized orbitals of carbon

atoms. An electron in the pz-orbital of each sp2-hybridized carbon atom will form

π-bonds with neighboring pz electrons. In conjugated polymers with chains of

alternating single–double bond structure, the molecular pz orbitals constituting

the π-bonds are actually overlapped and spread over the entire molecule. The

electrons in this molecular orbital are respectively delocalized along the whole

molecular chain, resulting in high electron polarizability.

Organic semiconducting polymers have a few distinct features, which distin-

guish them from their crystalline inorganic counterparts. First, organic materials

have poor carrier mobilities, typically less than 1 cm2/V ·s in most materials, which

is orders of magnitude lower than inorganic semiconductors. The highest mobility

reported to date is around 5 cm2/V · s in the pentacene-based field effect transistor

through a surface modification [98], and yet this is still two orders of magnitude

lower than silicon. Second, most organic semiconductors have absorption peaks

located in the visible range with strong absorption coefficients typically greater

than 105 cm−1. In most practical devices, however, this high extinction requires

an active layer thicknesses of only a few hundred nanometers, which partly com-

pensates for the low mobilities. Third, the absorption of incident photons in these

materials lead to photo-excitation of excitons (referring to mobile excited states or

118

S

S

S

S

*

*n

P3HT

PCBM-C60

e-

hν

OMe

O

Figure 5.1: Schematic drawing illustrating photoinduced charge transfer between a

conjugated polymer (P3HT) and a fullerene (PCBM-C60). After photo-excitation

P3HT, the electron is transferred to the PCBM-C60 within 10−13 second. With

respect to electron transfer, they are often referred to as donor and acceptor ma-

terials. In this particular example, P3HT is an electron donor and fullerene is an

electron acceptor material.

bound electron-hole pairs via the Coulomb force) rather than free charge carriers.

These excitons, an intermediate product in the photocurrent generation process,

carry energy but no net charge. Hence they diffuse within the material as a single

entity and recombine within distances, typically on the order of 10 nm [99–101].

In solar cells or photodetectors, they need to be dissociated non-radiatively by an

applied electric field exceeding the exciton binding energy, which can be up to 2 eV,

or through a process called photoinduced charge transfer [102–105], after which

free carriers can be collected at the respective electrodes.

119

Several approaches of active layer composites for organic PV cells or detectors

have been made over the last few decades. The main concepts of these approaches

are fundamental photophysics based on the photoinduced charge transfer either be-

tween layers of conjugated polymers and low molecular weight organic molecules

forming a single heterojunction at the interface, or within a single layer blend

of conjugated polymers and low molecular weight organic molecules forming a

bulk heterojunction. The very first generation of organic thin film PV cells was

based on a single organic layer sandwiched between two metal electrodes of dif-

ferent work functions [88], but the external quantum efficiency (EQE) and hence

the power conversion efficiency (PCE) was very poor, generally ≪ 1%. The re-

markable breakthrough was achieved by introducing bilayer heterojunction and

bulk heterojunction devices consisting of buckminster fullerene C60 molecules or

soluble derivatives of C60 in addition to the active conjugated polymers. After

photo-excitation of the conjugated polymer by an incident light with an energy ~ω

greater than the π-π∗ gap, electrons are transferred onto the fullerene molecules due

to their high electron affinity. Because an electron is transferred from the p-type

hole conducting polymer to the n-type electron conducting C60, these materials are

often referred to as electron donors and electron acceptors, respectively. Fig. 5.2(a)

illustrates a schematic energy band diagram of electrodes with their work-functions,

and donor/acceptor materials with their ionization potential Ip, electron affinity

χ, and energy band-gap Eg between HOMO (highest occupied molecular orbital)

and LUMO (lowest unoccupied molecular orbital) levels corresponding to the va-

120

lence and conduction bands in inorganic semiconductors, respectively. For efficient

charge transfer, an exciton binding energy greater than IDp −χA

p is required because

it is energetically favorable. Detailed time-resolved studies on this photoinduced

charge transfer between conjugated polymers and fullerene molecules showed that

it occurs within 10−13 second after photo-excitation of the polymer, which is nearly

103 faster than any other competing process [106]. For this reason, the charge

transfer efficiency approaches 100%, resulting in excellent quenching of excitonic

photoluminescence from the polymer.

Figure 5.3 depicts conceptually the photoinduced charge transfer in bilayer and

bulk heterojunction structures. In the bilayer structure, where the electron donor

(p-type hole conducting polymer) and electron acceptor (n-type electron conduct-

ing fullerene) form a well-defined planar interface by sequential deposition or spin-

casting of the organic layers, the EQE and PCE are limited primarily by the exciton

lifetime before photoexcited excitons recombine or dissociate into free charge carri-

ers at the D-A interface. Therefore, the device thickness should be kept within the

exciton diffusion length of the active polymer material for efficient charge trans-

fer. On the other hand, in the bulk heterojunction device, the D-A interface is

distributed over the entire active layer structure because both donor and acceptor

materials are co-evaporated or blended in solution. Therefore, the interfacial area

is increased by a large extent such that the dissociation site at a D-A interface falls

within a distance closer than the exciton diffusion length from an each absorbing

site. For this reason, the charge generation efficiency in the bulk heterojunction

121

hν

+

−

−

+

Ebi

EF1

IPD

IPD

IPA

IPA

EF2

χA

χA

χD

χD

EgD

EgD

EgD

EgD

(a)

(b)

ηA

ηED

ηCT

ηCC

+

−

Figure 5.2: (a) Schematic energy level diagram of electrodes with Fermi energy,

and a donor and acceptor materials with energy band-gap Eg between HOMO and

LUMO, ionization potential Ip and electron affinity χ. For efficient charge transfer,

an exciton binding energy greater than IDp −χA is required. (b) Schematic illustra-

tion of photocurrent generation in heterojunction PV cells and PDs. Generation of

photocurrent consists of four steps: exciton generation, exciton diffusion, exciton

dissociation by photoinduced charge transfer, and free charge carrier collection at

the electrodes. Therefore, the external quantum efficiency, the ratio of the number

of free charge carriers to the number of incident photons, can be expressed with

four respectively corresponding efficiencies, ηA, ηED, ηCT, ηCC.

122

structure is known to be near 100%. However, in such a blended structure, effi-

cient charge transport to the electrodes is sensitive to the nanoscale morphology

of the mixture because it requires an intimate interpenetrating network to pro-

vide percolated pathways for the hole and electron conducting paths to reach the

opposite contacts. Nanomorphology is dependent on many parameters: an evap-

oration rate, substrate temperature during vacuum deposition process, specific

solvents used and solvent evaporation time for solution process like spin-casting,

and the annealing condition and relative concentration of the mixture for both

processes [90]. It should be noted that charge transport in actual devices can be

much more complicated than the continuous conducting paths as illustrated in

Fig. 5.3(b) [107].

Photocurrent generation in heterojunction PV cells and PDs consists of four

steps and the corresponding efficiencies are defined as:

1. Exciton generation in the electron donor material from incident photons,

which is directly related to the material absorption coefficient, (ηA).

2. Exciton diffusion, fraction of excitons generated in the donor material reach-

ing the D-A interface, (ηED).

3. Exciton dissociation at the D-A interface through photoinduced charge trans-

fer, (ηCT).

4. Charge collection at the electrodes, (ηCC).

123

-+

+

-

+

- -

+

OMe

O

S

S

S

S* *n

(a) Bylayer heterojunction

(b) Bulk heterojunction

Figure 5.3: Schematic diagram depicting conceptually the photoinduced charge

transfer in (a)bilayer heterojunction and (b)bulk heterojunction. In a bilayer struc-

ture, the electron donor and electron acceptor form a well-defined planar interface

by sequential deposition or spin-casting. On the other hand, in a bulk heterojunc-

tion, the D-A interface is distributed over the entire active layer structure and its

interfacial area is increased by a large extent because both the donor and acceptor

materials are co-evaporated or blended in solution.

124

Therefore, we can write the external quantum efficiency ηEQE, which describes the

ratio of the number of free charge carriers to the number of incident photons

ηEQE =Iph/e

Plight/(hν)

=Nex

Nph

· NDAex

Nex

· NDAfree

NDAex

· NA,Cfree

NDAfree

= ηA · ηED · ηCT · ηCC

= ηA · ηIQE, (5.1)

where ηIQE = ηED · ηCT · ηCC is the internal quantum efficiency (IQE).

In this chapter, we will focus on the bulk heterojunction structure to develop

a photodetector with a high EQE. Although there are a number of different or-

ganic materials used in photovoltaic applications, including organic dyes or pig-

ments, the most commonly used active materials are, for hole conducting donor

materials, MDMO-PPV (poly[2-methoxy-5-(3,7-dimethyloctyloxy)]-1,4-phenylene-

vinylene), MEH-PPV (poly(2-methoxy-5-(2′-ethylhexyloxy)-1,4-phenylenevinylene)),

P3HT (poly(3-hexylthiophene-2,5-diyl)), P3OT (poly(3-octylthiophene)), PFB (poly-

(9,9′-dioctylfluorene-co-bis-N,N ′-(4-butylphenyl)-bis-N,N ′- phenyl-1,4-phenylen di-

amine), and pentacene, and for electron conducting acceptor materials, CN-PPV

(poly(2,5,2′,5′-tetrahexyloxy-7,8′-dicyano-di-p-phenylenevinylene), CN-MEH-PPV

(poly[2-methoxy-5-(2′-ethylhexyloxy)-1,4-(1-cyanovinylene)-phenylene), F8TB (poly-

(9,9′- dioctylfluoreneco-benzothiadiazole), fullerene C60 and C70, and chemically

modified soluble fullerene derivatives, namely PCBM-C60 ([6,6]-phenyl-C61-butyric

acid methyl ester) and PCBM-C70 ([6,6]-phenyl-C71-butyric acid methyl ester).

125

The materials used in this work are conjugated polymers, MEH-PPV and P3HT

for electron donors, and small organic molecules, PCBM-C60 and PCBM-C70 for

electron acceptor materials. Fig. 5.4 shows the chemical structures of these four

materials.

5.2 Fabrication of bulk heterojunction PDs

The device structure is shown in Fig. 5.5. Indium-tin-oxide (ITO) on float glass

substrate was purchased from Delta Technologies. The transparent conducting

oxide, ITO, with a sheet resistance of 4−8 Ω/ and a thickness of 200 nm was used

as the hole injecting electrode. In order to etch parts of the ITO to define several

devices on a 1′′ × 1′′ ITO/glass substrate, a photoresist pattern was transferred

onto the ITO layer by wet chemical etching using diluted HCl2. The etch rate can

be adjusted by varying the concentration of HCl. The typical etch rate is about

50 A/min in HCl aqueous solution (1:1, standard 38% HCl to DI water) at room

temperature and can be accelerated up to 2000 A/min in undiluted HCl (38%).

A digital voltmeter was used to check whether the ITO was completely cleared

without leaving any unwanted residual layer that can short out the device.

Once the patterned ITO/glass substrate was prepared, all the subsequent layers

were deposited in an inert N2 environment in a glove box to avoid photo-oxidative

degradation. After solvent cleaning of the ITO/glass substrate, a thin layer (∼

2ITO can be easily removed in HF as well. However, the etch rate is very high and often

uncontrollable, and, more importantly, it attacks the glass substrate at the same time. Reactive

ion etching in an Ar plasma is an alternative way to etch ITO when the feature size is small.

126

O

MeO

CH3

CH3

CH3

CH3

n

S

S

S

S* *n

(a) MEH-PPV

(b) P3HT

(c) PCBM-C60 (d) PCBM-C

70

OMe

O

OMe

O

Figure 5.4: The chemical structure of the materials in bulk heterojunction layer for

organic photodetector. (a) and (b) are electron donor/hole conducting conjugated

polymers, MEH-PPV and P3HT, respectively. (c) and (d) are electron accep-

tor/electron conducting small molecules, PCBM-C60 and PCBM-C70, respectively.

127

Glass substrate

ITO

PEDOT:PSS

Active layer

Al LiF

Figure 5.5: Schematic diagram of the device configuration.

50 nm) of PEDOT:PSS (poly[3,4-ethylenedioxythiophene]:poly[styrenesulfonate],

LED grade PEDOT Al4083 purchased from Baytron) was spin-coated and followed

by annealing at 150 C for 1 hr. Either ITO or PEDOT:PSS alone can be used

as the hole injecting anode, but the ITO/PEDOT:PSS combination employed in

OLEDs and OPV cells has given the most promising results for device efficiency

as well as lifetime [108–110].

A polymer/fullerene blend was spin-casted and annealed at 100 C for 1 hr for

the active layer with a desired thickness. The materials for the blend are conjugated

polymers MEH-PPV and P3HT for electron donors, and small organic molecules

PCBM-C60 and PCBM-C70 for electron acceptor materials, all of which were pur-

chased from American Dye Source. Chlorobenzene was used to dissolve both of

these organic materials for a solution process because chlorobenzene can provide

more uniform mixing of the constituents compared with other organic solvents

128

such as toluene, which is also frequently used. This more uniform mixing mainly

comes from the higher solubility of C60, and results in a smooth surface morphol-

ogy and hence an improved photocurrent [111]. Separate solutions of polymer and

fullerene were prepared and stirred at 80 C for a few days and then filtered into a

pre-cleaned vial to prepare a blend solution with a desired relative concentration

of donor/acceptor. The total solution concentration in weight was kept about 1%

for a layer with a thickness about 100 nm.

As mentioned earlier, the nanoscale morphology of the active layer in a bulk-

heterojunction is an important factor because the presence of an intimate inter-

penetrating network between the electron donor and acceptor materials is critical

for efficient charge transport. For this reason, a detailed study was conducted

earlier in our group (R. Barber, W. Herman, and D. Romero) to investigate how

the morphology and device characteristics depend on the relative concentration of

the active layer blend [112,113].

Figure 5.6 shows the atomic force microscopy (AFM) images for four different

PCBM-C60 molar fractions in the blend with a block copolymer MEH-PPV-co-

biphenylene vinylene [113]. The molar fraction of PCBM-C60 x is defined as

x =WC60

/MWC60

WC60/MWC60

+ Wpoly/MWpoly

. (5.2)

Therefore, x = 1 denotes pure C60, x = 0 denotes pure MEH-PPV polymer, and

x = 0.5 corresponds to the blend containing one C60 molecule for every repeat-

ing unit of the polymer. The AFM images of different surface morphologies with

129

5 µm 5 µm

5 µm 5 µm

c) x=0.26 d) x=0

a) x=1 b) x=0.43

Figure 5.6: Atomic Force Microscopy (AFM) images for four different PCBM-C60

molar fractions x [113].

130

varying fullerene concentrations clearly show that smooth morphology with uni-

form fine grains (∼ 100 nm), rather than segregated large clusters (∼ 1 µm) can

be achieved at intermediate values of x.

Further study on photovoltaic device characteristics depending on fullerene

concentration showed that when the molar fraction of C60 in the MEH-PPV

copolymer/C60 blend is around 0.6 (or equivalently, 3:1 in weight for this par-

ticular example), the device exhibits the best performance in terms of short circuit

current, fill factor, EQE, and PCE [113]. Note that the optimized relative ratio can

be significantly different when other materials are used. For instance in P3HT/C60

blend, the optimized ratio is close to x = 0.17 (or equivalently, 1.1:1 in weight for

this particular example).

For an electron-injecting top contact, a ∼ 1 nm thick layer of lithium-fluoride

(LiF) followed by a ∼ 60 nm thick layer of aluminum (Al) was deposited on top

of the organic layer via shadow mask technique using a thermal evaporator inside

the glove box. Similar to the case of the ITO/PEDOT:PSS combination that was

used for the hole-injecting electrode for the purpose of energy barrier engineering,

a very thin layer of LiF was inserted between the active layer and the Al cathode

in order to enhance the electron injection efficiency [114,115].

5.3 Device characteristics

Characterization of the devices was performed under an AM 1.5 solar simulator

with an intensity of 92 mW/cm2 and/or using a CW laser at a wavelength of 532 nm

131

with a maximum output power of 5 W.

A practical figure of merit for PV cells is the PCE (ηP), which is defined as the

ratio of the maximum electrical power generated to the incident light power [116],

ηP =maxI · V

Plight

=ISC · VOC

Plight

FF, (5.3)

where maxI · V is the maximum electrical power that the device can deliver to

an external load, Plight is the incident light power, ISC is the short-circuit current

(current at zero voltage), VOC is the open-circuit voltage (voltage intercept at zero

current), and FF is the fill factor defined as

FF =maxI · V ISC · VOC

, (5.4)

The PCE can also be expressed with the EQE (ηEQE) and the energy conversion

efficiency (ξen) at a particular wavelength as

ηP = ηEQE · ξen · FF, (5.5)

with

ηEQE =Iph/e

Plight/(hν), (5.6)

ξen =eVOC

hν. (5.7)

Note, however, that Eqs. 5.5–5.7 are more useful for describing the detector re-

sponse at a particular wavelength because PV cell are characterized under AM

132

-1.0 -0.5 0.0 0.5 1.0-8

-4

0

4

8

JS

C

(m

A/c

m-2 )

Bias voltage (V)

dark

AM 1.5

Figure 5.7: Current versus voltage (I-V ) characteristics of the bulk heterojunction

device of MEH-PPV copolymer and PCBM-C60 with a C60 molar fraction x = 0.57.

VOC = 0.88 V, JSC = 4.92 mA/cm2, FF = 0.47, and ηP = 2.4% under AM1.5 direct

solar simulator with Pin = 92 mW/cm2 [113].

1.5 solar illumination in most cases. The incident photon to current conversion

efficiency (IPCE),

IPCE[%] =1240 · ISC[µA]

λ[nm] · Plight[W ](5.8)

is often used instead of EQE (ηEQE), but they are equivalent to each other.

Figure 5.7 shows the current versus voltage (I-V ) characteristics in the dark and

under solar illumination for the bulk heterojunction device of MEH-PPV copoly-

mer and PCBM-C60 with a C60 molar fraction x = 0.57. We can extract the PV

133

characteristics of the device, VOC = 0.88 V, JSC = 4.92 mA/cm2, FF = 0.47, and

ηP = 2.4% under AM 1.5 direct solar simulator with Pin = 92 mW/cm2.

A practical figure of merit for PDs is the EQE or photo-responsivity at a certain

wavelength, while it is the PCE for the PV cells. In order to describe an overall

detector performance, a high EQE alone is not sufficient and other parameters

such as dark current, noise equivalent power, dynamic range, and response time,

should be considered as well. However, a high EQE is a prerequisite. Although a

detector can be used in ‘PV mode’, operating without any bias voltage applied,

most detectors operate under reverse bias voltage for fast carrier sweep within the

active layer. Therefore, depending on whether the device is employed as a solar

cell or a detector, a slightly different approach is needed. Our primary interest in

the rest of this chapter is to improve the EQE at the wavelengths of peak detector

responsivity by increasing the photocurrent under reverse bias.

Figure 5.9 shows the I-V characteristics of the bulk heterojunction detector of

P3HT/PCBM-C60 with a relative ratio, C60/P3HT = 1.15 in weight, (a) under AM

1.5 solar illumination and (b) under different laser intensities at 532 nm. As shown

in Fig. 5.8, the absorption peak of the blend of P3HT/PCBM-C60 is near 532 nm,

and hence the maximum detector responsivity is expected at around 532 nm as

well. Although the PCE under solar illumination is poor ηP < 0.5% mainly due to

the small fill factor FF ≈ 0.2, the EQE is around 0.3 at short circuit condition and

approaches 0.6 at V = −10 V as shown in the figure. These lower PCE but higher

EQE, in comparison with the previous device from MEH-PPV/C60 (ηp = 2.4 %

134

400 600 800 1000 12000.0

0.2

0.4

0.6

0.8

1.0

Absorb

ance

Wavelength (nm)

Figure 5.8: Absorption spectra of the blend of P3HT/PCBM-C60. The absorption

peak of the blend of P3HT/PCBM-C60 is near 532 nm.

and ηEQE < 0.1) reported in the earlier work [113], clearly indicate that an efficient

PV cell with a high PCE is not necessarily an efficient detector requiring a high

EQE and a slightly different approach is needed depends on whether the device is

employed as a solar cell or a detector. The value of EQE achieved (ηEQE = 0.3

and 0.6 at V = 0 V and −10 V, respectively) is reasonably high, but in order to

improve it further, we need to estimate how many photons arrive and are absorbed

within the active layer by taking into account of multilayer thin film interference

effects.

135

10 8 6 4 2 0 27

6

5

4

3

2

1

0

1

2

Bias voltage (V)

Photo

curr

ent densi

ty (

mA

/cm

2)

(a) under solar AM 1.5 illumination

10 8 6 4 2 0 22

1. 5

1

0. 5

0

0.5

1

Bias voltage (V)

Photo

curr

ent densi

ty (

mA

/cm

2)

(b) under laser (@ 532 nm) illumination

Figure 5.9: I-V characteristics of the bulk heterojunction device of P3HT and

PCBM-C60 with a C60/P3HT 1.1 in weight, (a) under AM 1.5 direct solar illumi-

nation (b) under different laser intensities at 532 nm.

136

-10 -8 -6 -4 -2 00

0.2

0.4

0.6

0.8

1

Bias volatge (V)

Ext

ern

al q

uantu

m e

ffic

iency

,

η

EX

E

P=2.06 mW/cm2

P=3.13 mW/cm2

P=4.92 mW/cm2

P=5.78 mW/cm2

P=6.95 mW/cm2

Figure 5.10: External quantum efficiency vs. bias voltage of the device from P3HT

and PCBM-C60 at 532 nm. ηEQE is around 0.3 at short circuit condition and

approaches 0.6 at V = −10 V.

137

5.4 Multilayer thin film interference effects

Because the number of excitons generated is directly related to the absorption of

optical energy, in order to improve the external quantum efficiency, the first step

to consider is maximizing the number of photons that arrive in the active layer,

where photoexcited excitons are generated and dissociated into free carriers.

Because the exciton diffusion and dissociation efficiencies are near 100% in the

bulk heterojunction as mentioned in section 5.1, we can define a theoretical upper

limit of EQE that assumes all the free carriers from the dissociated excitons are

collected at the respective electrodes (ηCC = 1). We expect that this upper limit of

EQE can be nearly achieved when the detector is under a relatively strong reverse

bias.

For the estimation of EQE, a model including interference effects from multiple

reflection and transmission at each interface is necessary to calculate the optical

field distribution in each layer. This interference effect is particularly important

when there is a very strong reflection at the conducting electrode and polymer

interface as the case of our device. When considering a device model of a stratified

structure, knowledge of the complex refractive indices and thicknesses of all the

layers is a prerequisite.

5.4.1 Optical properties of individual layers

For the numerical modeling to calculate the optical field profile in the stratified

structure, the complex refractive indices and thicknesses of individual layers were

138

determined by a spectroscopic ellipsometry (J.A. Wollam) employing a model fit-

ting procedure using the Drude-Lorentz model [117, 118]. The complex dielectric

function ǫ(ω) in the Drude-Lorentz model can be expressed as

ǫ(ω) = ǫR − jǫI (5.9)

= ǫ∞ +f0ω

2p

−ω2 + jωΓ0

+n

∑

k=1

fkω2p

(ω2k − ω2) + jωΓk

, . (5.10)

where ωp is the plasma frequency and n is the number of Lorentz oscillators with

a center frequency of ωk, a strength of fk and a damping constant Γk. This Drude-

Lorentz model combines the free-electron model to represent intraband optical

transition and the bound-electron model for interband effects. In general, there

can be one or more Lorentz oscillators depending on the material. The complex

index of refraction n can be determined from the complex dielectric function:

n =√

ǫ = n − jk. (5.11)

Figure 5.11 (a) shows the measurement result for the real and imaginary index

of refraction of the ITO on soda lime glass substrate. For the model fit to extract n

and k from ellipsometric data, one Lorentz oscillator in the Drude-Lorentz model

(n = 1 in Eq. 5.10) was enough to achieve an excellent fit in the case of ITO [119].

Also note that the thickness obtained from a profilometer was used for an initial

guess value in the fitting procedure.

In order to validate the measurement of the complex refractive index, the values

of n and k, and the thickness from the best-fit model for the ellipsometric data

139

were used to generate the transmission data, and the generated transmission curve

was compared with the experimentally measured UV/Vis/IR transmission from a

spectrophotometer as shown in Fig. 5.11 (b).

With the same procedure as ITO, the optical properties and thicknesses of other

layers were also determined. Figure 5.12 shows the n and k for Al, PEDOT:PSS,

and P3HT:C60. The optical properties of LiF were not shown in the figure because

the index dispersion of LiF is very flat, ∼ 1.39, in both the visible and near-IR

region, and absorption in a 1 nm thick LiF layer is negligible.

5.4.2 Effect of multilayer thin film interference

Once all the optical properties and thicknesses of individual layers were determined

as discussed in the previous section, we can calculate the optical field distribution

and absorption of light in each layer of the stratified device structure. Although

there are a few different ways to perform the calculation, we follow the approach

of the work reported in [91, 101]. In this model, we assume a plane wave incident

normal to the device, homogeneous and isotropic layers, and parallel and smooth

interfaces.

To calculate the optical electric field as a function of position within the device,

consider the schematic diagram of the multilayer structure of the device as shown

in Fig. 5.13. The field in the i-th layer has two components, E+i and E−

i , propa-

gating in the positive and the negative z direction, respectively. In this case, the

electric field of the light can be represented by a 2 × 2 matrix because equations

140

300 500 700 900 1100 1300 15000

0.5

1

1.5

2

2.5

Wavelength (nm)

n

300 500 700 900 1100 1300 15000

0.5

1

1.5

2

2.5

k

n

k

(a) n and k of ITO film

400 600 800 1000 1200 14000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Wavelength (nm)

Tra

nsm

issi

on

modelUV/Vis/IR

Substrate

R

T

Film

(b) Transmission of ITO/glass sample

Figure 5.11: (a) Real and imaginary index of refraction measured from the spec-

troscopic ellipsometric data and model fit with the Drude-Lorentz model. (b)

Comparison between the estimated transmission using the n and k, and thickness

from the best-fit model and the measured transmission from a spectrophotometer.

141

300 400 500 600 700 800 900 10000

0.5

1

1.5

2

2.5

3

Wavelength (nm)

n

AlPEDOTP3HT:C

60

(a)

300 400 500 600 700 800 900 10000

1

2

3

4

5

6

7

8

9

10

Wavelength (nm)

k

AlPEDOTP3HT:C

60

(b)

Figure 5.12: Optical properties of Al, PEDOT:PSS, and P3HT:C60

142

Glass Substrate ITO PEDOT:PSS

Active layer

AlLiF

Rs

TsR

T

RD

TD

d0=1 mm d1=200 nm d2=50 nm d3=dA d4=1 nm d5=60 nm

E0+ E1

+ E2+ E3

+ E4+ E5

+ E6+

E0- E1

- E2- E3

- E4- E5

- E6-

xz

y

Figure 5.13: Schematic diagram of the multilayer device structure with the thick-

nesses and the optical electric fields in each layer, E+i and E−

i , representing the

component propagating in the positive and the negative z direction, respectively.

describing the light propagation are linear, and the tangential components of the

field are continuous across the interface. In this particular example, five thin film

layers are stacked to form the actual device. A 1 mm thick glass substrate was

not directly included in the thin-film analysis because the light travelling within

the glass substrate could be treated as effectively incoherent light by considering

the average effect combining the angular dispersion of incident light, the finite

bandwidth of the light source, and the surface roughness and nonuniformity in the

thickness of the substrate. The substrate effect can be incorporated by considering

the reflectivity RD and transmissivity TD of the entire device after we calculate R

143

and T of the multilayer structure, i.e.

RD =RS + R − 2RSR

1 − RSR, (5.12)

TD =TST

1 − RSR, (5.13)

AD = 1 − RD − TD, (5.14)

with

RS =

∣

∣

∣

∣

1 − n0

1 + n0

∣

∣

∣

∣

2

, (5.15)

TS = n0

∣

∣

∣

∣

2

1 + n0

∣

∣

∣

∣

2

, (5.16)

where RS and TS are respectively the reflectivity and transmissivity at the air/glass

interface, AD is the total absorption in the device, and n0 is the complex refractive

index of the glass substrate.

To calculate the optical fields as a function of position in the most general case,

consider m layers and corresponding fields Ei (i = 0 ...m + 1). Then we can write

an expression that relates the two outermost fields E0 and Em+1,

E+0

E−0

= S ·

E+m+1

E−m+1

, (5.17)

with the scattering matrix S defined as

S =

S11 S12

S21 S22

=

m∏

p=1

I(p−1)pLp

· Im(m+1), (5.18)

144

where Iij is the interface matrix representing the reflection and transmission at the

i-j interface and Li is the layer matrix representing a phase change through the

i-th layer as defined below.

Li =

ejk0nidi 0

0 e−jk0nidi

, (5.19)

Iij =

1tij

rij

tij

rij

tij

1tij

, (5.20)

rij =ni − nj

ni + nj

, (5.21)

tij =2ni

ni + nj

, (5.22)

where ni = ni − jki is the complex refractive index and di is the thickness of the

i-th layer. Note that the complex Fresnel reflection coefficient rij and transmission

coefficient tij are the same for TE and TM modes when the light is incident normal

to the surface.

When light is incident on the glass and travels through the device in the +x

direction and E−m+1 = 0, we can obtain the complex coefficients r and t.

r =E−

0

E+0

=S21

S11

, (5.23)

t =E+

m+1

E+0

=1

S11

, (5.24)

Therefore, the reflectivity R and transmissivity T described in Fig. 5.13 can be

145

written as

R = |r|2, (5.25)

T =|t|2n0

. (5.26)

To obtain the electric field within the i-th layer, the total multilayer transfer

matrix can be divided into two subsets, SLi and SR

i . Hence, we can write an

expression

S = SLi LiS

Ri , (5.27)

with

SLi =

i−1∏

p=1

I(p−1)pLp

· I(i−1)i, (5.28)

SRi =

m∏

p=i+1

I(p−1)pLp

· Im(m+1). (5.29)

From the equations above, we can obtain the electric field propagating in the +x

direction in the i-th layer with respect to the incident optical field,

E+i = t+i E+

0 =

1S−

i11

1 +S−

i12S+

i21

S−

i11S+

i11

e−j2k0nidi

E+0 , (5.30)

Similarly the electric field propagating in the −x direction in the i-th layer with

respect to the incident optical field is given by,

E−i = t−i E+

0 =

S+i21

S−

i11S+

i11

e−j2k0nidi

1 +S−

i12S+

i21

S−

i11S+

i11

e−j2k0nidi

E+0 . (5.31)

Therefore, the total electric field in the i-th layer at an arbitrary distance x to the

146

300 400 500 600 700 800 900 10000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

X: 532Y: 0.8322

Wavelength (nm)

RD, A

D

X: 532Y: 0.1634

Absorption (1-RD -TD )

Reflection (RD)

Glass

Figure 5.14: Calculation of reflectivity RD and absorption AD = 1−RD−TD of the

device illustrated in Fig. 5.13 with a bulk heterojunction active layer P3HT/C60

with a thickness 100 nm.

right of the (i-1)/i interface can be written as

Ei(x) = E+i (x) + E−

i (x) (5.32)

= (t+i e−jk0nidi + t−i ejk0nidi)E+0 . (5.33)

Figure 5.14 depicts the calculated reflectivity RD and absorption AD = 1 −

RD − TD of the bulk heterojunction device from P3HT/C60 with a thickness of

100 nm. Interestingly, at a wavelength of 532 nm, about 16.3% of the total inci-

dent photons are reflected at the air/device interface and do not even reach the

active layer to generate excitons. This reflectivity is much higher than the 4%

147

50 100 150 200 250 300 350 4000

0.2

0.4

0.6

0.8

1

Distance from sub/ITO interface (nm)

|E(x

)|2

ITO AlPEDOT:PSSPEDOT:PSS LiFLiFP3HT/C60

Figure 5.15: Calculation of field distribution as a function of position. Modulus

squared of the optical electric field |E(x)|2 is normalized to the incident light field

|E+0 |2.

reflection at a simple air/glass interface, which might have been expected. This

calculation directly indicates how strongly the multilayer interference depends on

how we construct the practical device structure. It also clearly states that the

external quantum efficiency – the fraction of incident photons in the device that

can contribute to the photocurrent – can never exceed 83.2%, which is the total

light absorption in the entire device. In other words, if all the absorbed photons

could contribute to the photocurrent, an EQE of 83.2% would be obtained.

Figure 5.15 shows the calculated field distribution as a function of position. The

modulus squared of the optical electric field |E(x)|2 is normalized to the incident

148

light field |E+0 |2.

It is worth pointing out that a detailed field profile within the active layer

is very crucial in a bilayer structure because, in order to maximize the number

of excitons at the donor/acceptor interface, where the excitons dissociate most

efficiently into free carriers, a global peak of |E(x)|2 needs to be located near the

interface [101]. However, this is not the primary factor determining the EQE in

the bulk heterojunction device.

Once the optical field profile is found, we can calculate the time-averaged ab-

sorbed power or the number of absorbed photons as a function of position, which

is directly proportional to the number of excitons generated.

Qi(x) = αiIi(x)

=1

2cǫ0niαi|Ei(x)|2

=2πcǫ0niki

λ|Ei(x)|2, (5.34)

where αi is the absorption coefficient, and Ii(x) is the light intensity in the i-th

layer.

Figure 5.16 plots the calculated absorbed power Q(x) as a function of position.

Note that the simple exponential model Q(x) = αI0 exp (−αx), neglecting multiple

reflections and transmissions at the interfaces between the layers, can only give an

approximate exciton generation profile [120,121]. The total absorbed light within

the active layer only is given by

IAabs =

∫

dA

QA(x) dx, (5.35)

149

50 100 150 200 250 300 350 4000

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2x 10

4

Distance from sub/ITO interface (nm)

Q(x

)

ITO AlPEDOT:PSSPEDOT:PSS LiFLiFP3HT/C60

Figure 5.16: Calculation of time-averaged absorbed power Q(x) as a function of

position.

and we can define the upper limit to the external quantum efficiency ηLEQE as

ηLEQE =

IAabs

Iinc

, (5.36)

where Iinc is the intensity of the incident light. This ηLEQE represents the theoretical

upper limit of EQE that can be achieved from a specific device structure once all

the optical properties and thicknesses of individual layers are measured. In other

words, we can achieve ηLEQE, the maximum incident photons to current conversion

efficiency, when all the photons absorbed in the active layer contribute to the

photocurrent. This is equivalent to assuming that the internal quantum efficiency

ηIQE is unity, i.e. the exciton diffusion ηED, exciton dissociation ηCT, and charge

150

collection efficiency ηCC in Eq. 5.6 are 100%.

It should be noted that, for a comprehensive theoretical estimation of the EQE,

we need to solve the one-dimensional steady-state carrier transport equation and

current density equation [122,123].

Dd2

dx2n(x) + µE

d

dxn(x) − n(x)

τ+

γQ(x)

hν= 0 (5.37)

J(x) = qµEn(x) + qDd

dxn(x) (5.38)

where D is the diffusion constant, µ is the charge mobility, E is electric field, n(x)

is the carrier density, τ is the carrier life-time, and γ is the carrier generation rate,

which is equivalent to ηED ·ηCT. However, it is not straightforward to solve this car-

rier transport equation numerically, mainly because it is a multi-point boundary

value problem requiring all the boundary conditions at the interfaces. In addi-

tion, the field-dependence of the charge mobility cannot be neglected under strong

electric fields associated with the applied reverse bias. Therefore determining the

charge mobility becomes very crucial in the calculation.

Nevertheless, the ηLEQE is still a very useful quantity for the photodetector under

applied fields E ≫ 106V/m because, under such strong fields, ηCC can easily reach

80–90%, and this leads to a value of EQE about 80–90% of ηLEQE.

The calculated ηLEQE for the device structure above is 0.69. Therefore, the EQE

we obtained (ηEQE = 0.6 at −10 V as shown in Fig. 5.10) is 87% of ηLEQE, leading

to ηCC = 0.87.

151

5.5 Improvement of the external quantum efficiency

So far, we used PEDOT:PSS and LiF together with ITO and Al, respectively, to

form electrodes for more efficient hole and electron injection. This is a desirable

approach to achieve high short-circuit currents and high PCE in PV cells by com-

pensating energy level offset or band bending at non-ideal ohmic contacts. In the

detector, however, the carriers are expected to have enough energy to overcome

these energy level misalignments under a strong applied reverse bias. In addition,

the PEDOT:PSS layer absorbs a non-negligible amount of photons and reduces

the number of photons absorbed within the active layer.

Figure 5.17 shows the calculated field profile |E|2 and time-averaged absorbed

power Q as a function of position in the case where there is no PEDOT:PSS and

LiF. As we might have expected, the upper limit of the external quantum effi-

ciency ηLEQE increases to 87.8% from 69% in the previous case of glass/ ITO/ PE-

DOT:PSS/ P3HT:C60/ LiF/ Al.

In order to study the active layer thickness dA dependence on ηEQE, we fabri-

cated devices with three different thicknesses shown in Fig. 5.18 with a structure

of glass/ITO/P3HT:C60/Al. Figure 5.19 plots the calculation of ηLEQE with (dotted

line) and without PEDOT:PSS and LiF (solid line) and measured EQE at -10 V

(triangles). Interestingly, when PEDOT:PSS and LiF are used, the EQE has a

global peak value of 0.69 at a thickness of 100 nm and has a local minimum of 0.60

at a thickness of about 170 nm, which can be explained by multilayer interference

152

50 100 150 200 250 300 3500

0.2

0.4

0.6

0.8

1

Distance from sub/ITO interface (nm)

|E(x

)|2

ITO AlP3HT/C60

(a)

50 100 150 200 250 300 3500

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2x 10

4

Distance from sub/ITO interface (nm)

Q(x

)

ITO AlP3HT/C60

(b)

Figure 5.17: Calculation of field distribution |E(x)|2 and time-averaged ab-

sorbed power Q(x) as a function of position for the device structure of

glass/ITO/P3HT:C60/Al.

153

400 600 800 10000

20

40

60

80

100

120

300 nm

120 nm

100 nm

Tra

nsm

issio

n

Wavelength (nm)

Figure 5.18: UV-Vis-IR transmission spectra for P3HT:C60 samples with three

different thicknesses, 100 nm, 120 nm, and 300 nm,

as well. On the other hand, when PEDOT:PSS and LiF layers are eliminated, the

active layer thickness dependence is much less noticeable and ηLEQE becomes fairly

flat (0.87 − 0.90) for the thickness dA > 100 nm.

Figure 5.20 shows the measurement results of the EQE as a function of bias

voltage and photocurrent at V = −10 V as a function of laser intensity at 532 nm,

both of which were obtained from I-V characteristics of the device with a thick-

ness dA = 300 nm. The value of EQE at short circuit operation is negligibly small

because the PEDOT:PSS and LiF layers have been removed. In this case the

thickness is considered to be quite large compared to other devices resulting in a

small electric field across the device. However, we are primarily interested in the

154

0 50 100 150 200 250 3000

0.2

0.4

0.6

0.8

1

Active layer thickness (nm)

ηE

QE

L

Figure 5.19: Calculated upper limit of the external quantum efficiency ηLEQE as

a function of active layer thickness dA (dotted and solid lines) and the measured

EQE at bias V =-10 V (triangles), with and without PEDOT:PSS and LiF layers.

case that it is under a strong electric field on the order of ∼ 107 V/m. Note that,

at V = −10 V, an External quantum efficiency ηEQE reaches 87±2%, leading to an

internal quantum efficiency ηIQE ≈ 97% in this particular example. This indicates

that an EQE can indeed reach near ηLEQE and hence a charge collection efficiency

across the intervening energy barriers can reach near 100% under a strong elec-

tric field. The achieved external quantum efficiency is one of the highest values

published so far. Previously, the highest reported values were ηEQE = 75% at

V = −10 V [91] and ηEQE = 78% at V = 0 V [94]. Unfortunately, the full scale

dynamic range of the detector could not be investigated due to the limited max-

imum intensity of the laser, but we did not observe any photocurrent saturation

155

up to an intensity of IL = 70 mW/cm2.

5.6 Summary

In this chapter, we investigated the highly efficient organic thin film bulk hetero-

junction photodetector fabricated from a blend of conjugated polymer (P3HT) and

small molecule (fullerene C60). We have benefited from the ideas developed in the

study of OPV cells because the OPD shares fundamental photophysics with the

OPV cell in terms of the photocurrent generation process. However, the detec-

tor needs a slightly different approach from that used in a photovoltaic cell. We

addressed the effect of multilayer thin film interference on the external quantum

efficiency. From the numerical modeling to calculate the optical field distribu-

tion and absorption of light energy in the layers, based on the characterization

of optical properties of individual layers comprising the detector, we proposed a

bulk heterojunction photodetector without PEDOT:PSS and LiF layers that are

commonly used in photovoltaic cells for high short-circuit current. Through the

experimental I-V characterization using a CW laser at a wavelength of 532 nm, we

demonstrated that it exhibited an external quantum efficiency ηEQE = 87±2% un-

der an applied bias voltage V = −10 V, leading to an internal quantum efficiency

ηIQE ≈ 97%. These results show that the charge collection efficiency across the in-

tervening energy barriers can indeed reach near 100% under a strong electric field.

The achieved external quantum efficiency is one of the highest values published so

far.

156

-10 -8 -6 -4 -2 00

0.2

0.4

0.6

0.8

1

Bias Volatge (V)

ηE

XT

(a)

0 10 20 30 40 50 60 700

0.1

0.2

0.3

0.4

0.5

Laser Intensity (mW/cm2)

Ph

oto

curr

ent

(mA

)

(b)

Figure 5.20: Measured ηEQE as a function of applied reverse bias voltage (a) and

photocurrent at V = −10 V as a function of incident laser intensity (b) for a device

with a dA = 300 nm.

157

Chapter 6

Conclusions

6.1 Summary and accomplishments

This dissertation work was motivated by emerging interests in investigating the

technological feasibilities of novel photonic and optoelectronic components fab-

ricated from new classes of organic materials. Due to this nature, the research

work was truly a multi-disciplinary subject and has been carried out with a strong

collaboration with chemists and physicists. The subject was grouped into three

principal parts describing different types of devices that are somewhat independent

of each other, but can be integrated into single devices; optical chiral waveguides

and magneto-optic materials, waveguides and microring devices from perfluori-

nated polymer, and organic bulk heterojunction photodetectors from conjugated

polymer and fullerene.

First, we studied unique polarization properties of novel organic chiral mate-

rials and optical chiral waveguides fabricated from binaphthyl-based chiral single

molecular compounds that can be formed into glassy isotropic thin solid films with

negligible birefringence. Asymmetric chiral-core planar waveguides were evaluated

for viable applications including an integrated-optic polarization rotator. Through

158

a detailed experimental polarization analysis on the waveguide eigenmodes, we

showed that the eigenmodes of the waveguides were indeed elliptically polarized

modes with the mode ellipticities on the order of 0.25, which agrees well with the

one predicted in the recent theory. The mode ellipticity that we obtained was not

large enough to meet the requirement for practical devices such as a polarization

rotator – predominant circularly polarized modes – but, to the best of our knowl-

edge, it was the first experimental demonstration of the mode ellipticities of the

chiral-core optical waveguides.

We also characterized novel organic magneto-optic materials, which can be vi-

able alternatives to the chiral materials. Measurement of the Verdet constants from

organic thin-film samples (10 − 100 µm) under a moderate magnetic field can be

very challenging and requires a very sensitive measurement technique. We adopted

a homodyne balanced detection with a sensitivity that can determine an angle of

polarization rotation as small as 0.5 × 10−6 radian and successfully measured the

Verdet constants of organic samples provided by Georgia Tech at different wave-

lengths. We measured Verdet constants of 10.4 and 4.2 rad/T · m at 1300 nm and

1550 nm, respectively, from an organic sample provided by Georgia Tech, which is

comparable to that of terbium gallium garnet. This unique observation can lead to

a new opportunity to build integrated-optic isolators or circulators from a simple

fabrication methodology including spin-casting and photolithography.

Second, we investigated low-loss waveguides and microring resonators fabri-

cated from perfluorocyclobutyl (PFCB) copolymer, which has substantially low

159

material absorption loss at telecommunication wavelengths. We discussed the de-

sign, fabrication, and characterization of those devices. We provided practical

design parameters through a waveguide eigenmode analysis, 3-D full vectorial cal-

culation for bending loss estimation, and the coupled mode theory for an estimation

of a coupling efficiency. A couple of fabrication challenges, an adhesion and mask

cracking problems, were also addressed. From the experimental characterization,

we demonstrated the two different types of straight waveguides, a buried channel

and pedestal structures, with propagation losses of 0.3 dB/cm and 1.1 dB/cm, re-

spectively. A microring add-drop filter with a maximum extinction ratio of 4.87 dB,

quality factor Q = 8554, and finesse F = 55 was also demonstrated. In addition,

we showed experimentally that all-optical switching with the PFCB microring res-

onator is possible when it is optically pumped with a femtosecond laser pulse of

sufficient energy. For a microring-loaded Mach-Zehnder interferometer (MR-MZI),

we demonstrated a modulation window of 30 ps and a maximum modulation depth

of 3.8 dB from an optical pump with a pulse duration of 100 fs and a pulse energy of

500 pJ when the signal wavelength is initially tuned to one of the ring resonances.

Finally, we investigated the highly efficient organic bulk heterojunction pho-

todetector fabricated from a blend of conjugated polymer and small molecule,

P3HT/PCBM-C60. For a numerical modelling accounting for multilayer interfer-

ence effects on the external quantum efficiency of a detector, the optical properties

and thicknesses of all the individual layers constituting the device were character-

ized by the spectroscopic ellipsometry and thin film transmission measurement.

160

Based on the numerical result, we proposed a bulk heterojunction photodetector

without containing PEDOT:PSS and LiF layers that are widely used for efficient

charge injection in a photovoltaic cell. Experimental I-V characterization at a

wavelength of 532 nm exhibited that the external quantum efficiency ηEQE could

reach near the theoretical upper limit efficiency ηLEQE under a strong electric field.

From a P3HT/C60 bulk heterojunction device, we achieved a very high external

quantum efficiency ηEQE = 87 ± 2% leading to an internal quantum efficiency

ηIQE ≈ 97% under an applied bias voltage V = −10 V. These results show that the

charge collection efficiency across the intervening energy barriers can indeed reach

near 100% under a strong electric field. The achieved external quantum efficiency

is one of the highest values obtained so far.

6.2 Future work

We have discussed the technological feasibility of photonic devices based on truly

novel organic materials. Therefore, by nature, the work discussed in this thesis

is still in progress in many aspects. Most of the materials that were used in this

dissertation work are not even commercially available and suffer from incomplete

understanding on their physical, optical and electronic properties. By the same

token, fabricating devices can be extremely challenging, and sometimes it can take

a very long time to develop a new process technique suitable for those materials.

Long term reliability is another problem that needs to be resolved for success-

ful commercialization. For these reason, a strong collaboration among engineers,

161

physicist, and chemist is required.

In order for chiral waveguides to be used as a TE/TM converter or a polar-

ization rotator, the eigenmodes must be predominantly circularly polarized states,

achievable by increased material chirality along with negligible birefringence. We

showed that, for a given material chirality, the mode ellipticity in a weekly guided

waveguide is much higher than a tightly confined waveguide. However, for nearly

circularly modes, a chirality at least an order magnitude larger (γ ≈ 4 pm or

equivalently, bulk ρ = 60 deg/mm) than the one we presented is required. It is

also desirable to extend the ORD maxima towards the telecommunication wave-

lengths for optical communication applications. Certain types of chiral compounds

are very sensitive to UV light and they tend to racemize in the air. Therefore, so-

lutions to this photoracemization need to be addressed as well in the future. It

should not be an easy task, but developing a numerical technique to analyze the

eigenmode and the mode ellipticity in 3D guiding structures can provide a very

useful tool for designing a practical integrated optic chiral devices.

The Verdet constants from organic materials we observed were moderate, but

this unique observation opens the possibility that integrated optic isolators or

circulators can easily be constructed from a simple fabrication methodology in-

cluding spin-casting and photolithography. However, in order to build practical

devices, much work needs to be done to synthesize the materials with even larger

Verdet constants, and to investigate the polarization properties of the eigenmodes

in magneto-optic waveguides. In addition, a comprehensive theoretical study in-

162

cluding a quantum mechanical approach is needed to understand the origin of the

large Verdet constants from organic materials.

More theoretical and experimental investigation is necessary for the PFCB-

based waveguide devices. Although we did not discuss the numerical techniques

to calculate the scattering loss, it can be done by modifying our finite difference

code properly to employ a radiative scattering source. Aside from achieving a

narrower gap, much needs to be done to refine the etch process in order to reduce

the scattering loss, which is a dominant source increasing the total loss in the

PFCB waveguides. A different etching condition in RIE, or different etching tools

including an inductive coupled plasma etching (ICP) can be tried. A post-etching

process such as chemical or thermal treatment may allow us to achieve much

smoother side wall. In addition, adhesion of fluorinated polymer to other layers

can be further improved by, for example, plasma surface treatment, and stress

arising from the CTE mismatch can be further eliminated. Other classes of novel

materials including electro-optic, semiconducting, and 3rd order nonlinear optical

polymer can be used to construct a ring resonator to exploit its multiple device

functionality.

Finally, for the development of organic photodetectors, other detector char-

acteristics such as dark current, noise equivalent power, dynamic range, response

time needs to be investigated. For long-term reliability, incorporating the encapsu-

lation technology developed for commercial OLEDs needs to be done for practical

use. A detector with a high photoresponsivity at infra-red wavelengths is worth

163

investigating. In particular, more advanced types of devices such as a travelling-

wave photodetector fabricated with a low loss polymer waveguide is certainly one

of the most interesting possible future developments.

I described several possible directions for future research, but potentially use-

ful devices from organic materials are limited only by the imagination. Didn’t I

mention that we are already living in the “PLASTIC” world? The future is about

to unfold.

164

BIBLIOGRAPHY

[1] L. Eldada, “Optical communication components,” Review of Scientific In-

struments, vol. 75, no. 3, pp. 575–593, 2004.

[2] H. Ma, K. Y. Jen, and L. R. Dalton, “Polymer-based optical waveguides:

Materials, processing, and devices.” Advanced Materials, vol. 14, no. 19, pp.

1339–1365, 2002.

[3] A. J. Heeger, “Nobel lecture: Semiconducting and metallic polymers: The

fourth generation of polymeric materials,” Reviews of Modern Physics,

vol. 73, no. 3, pp. 681–700, 2001.

[4] S. R. Forrest, “The path to ubiquitous and low-cost organic electronic appli-

ances on plastic,” Nature, vol. 428, no. 6986, pp. 911–918, 2004.

[5] J. McMurry, Organic Chemistry, 5th ed. Pacific Grove: Brooks/Cole, 1999.

[6] N. M. Davies and X. W. Teng, “Importance of chirality in drug therapy and

pharmacy practice: Implications for psychiatry,” Advances in Pharmacy,

vol. 1, no. 3, pp. 242–252, 2003.

[7] S. F. Mason, “Optical activity and molecular dissymmetry,” Contemporary

physics, vol. 9, no. 3, pp. 239–256, 1968.

165

[8] W. N. Herman, “Polarization eccentricity of the transverse field for modes

in chiral core planar waveguides,” Journal of Optical Society of America, A,

vol. 18, no. 11, pp. 2806–2818, 2001.

[9] C. Nuckolls, T. J. Katz, G. Katz, P. J. Collings, and L. Castellanos, “Synthe-

sis and aggregation of a conjugated helical molecule.” Journal of Americal

Chemical Society, vol. 121, no. 1, pp. 79–88, 1999.

[10] J. D. Jackson, Classical Electrodynamics, 3rd ed. New York: Jone Wiley &

Sons, 1999.

[11] M. P. Silverman, “Refection and refraction at the surface of a chiral medium:

comparison of gyrotropic constitutive relations invariant or noninvariant un-

der a duality transformation,” Journal of Optical Society of America, A,

vol. 3, no. 6, pp. 830–837, 1986.

[12] J. Lekner, “Optical properties of isotropic chiral media,” Pure and Applied

Optics, vol. 5, no. 4, pp. 417–443, 1996.

[13] I. V. Lindell, A. H. Sihvola, S. A. Tretyakov, and A. J. Vitanen, Electromag-

netic Waves in Chiral and Bi-Isotropic Media, Boston, 1994.

[14] D. Marcuse, Theory of Dielectric Optical Waveguides, 2nd ed. San Diego:

Academic Press, 1991.

166

[15] S. V. Demidov, K. V. Kushnarev, and V. V. Shevchenko, “Dispersion prop-

erties of the modes of chiral planar optical waveguides,” Journal of Commu-

nications Technology and Electronics, vol. 44, no. 7, pp. 827–832, 1999.

[16] M. Oksanen, P. K. Loivisto, and I. V. Lindell, “Dispersion curves and fields

for a chiral slab waveguide,” IEE Proceedings H, vol. 138, no. 4, pp. 327–334,

1991.

[17] F. Y. Fan, J. C. Mastrangelo, D. Katsis, S. H. Chen, and T. N. Blanton,

“Novel glass forming liquid crystals V. Nemetic and chiral-neumatic systems

with an elevated glass-transition temperature.” Liquid Crystals, vol. 27, no. 9,

pp. 1239–1248, 2000.

[18] J.-. W. Park, M. D. Ediger, and M. M. Green, “Chiral studies in amorphous

solids: The effect of the polymeric glassy state on the racemization kinetics

of bridged paddled binaphthyls.” Journal of Americal Chemical Society, vol.

123, no. 1, pp. 49–56, 2001.

[19] W. N. Herman, Y. Kim, W. L. Cao, J. Goldhar, C. H. Lee, M. M. Green,

V. Jain, and M. J. Lee, “Amorphous thin films of chiral binaphthyls for

photonic waveguides,” Journal of Macromolecular Science Part A - Pure

and Applied Chemistry, vol. A40, no. 12, pp. 1369–1382, 2003.

[20] G. D. Van Wiggeren and R. Roy, “High-speed fiber-optic polarization ana-

lyzer, measurements of the polarization dynamics of an erbium-doped fiber

167

ring laser,” Optical Communications, vol. 164, pp. 107–120, 1999.

[21] S. T. Tang and H. S. Kwok, “3×3 matrix for unitary optical systems,” Jour-

nal of Optical Society of America, A, vol. 18, no. 9, pp. 2138–2145, 2001.

[22] Y. P. Svirko and N. Zheludev, Polarization of Light in Nonlinear Optics.

New York: John Wiley & Sons, 1998.

[23] M. M. Green, “Personal communications.”

[24] M. Miyasaka, A. Rajca, M. Pink, and S. Rajca, “Chiral molecular glass: Syn-

thesis and characterization of enantiomerically pure thiophene-based [7]he-

licene,” Chemistry - A European Journal, vol. 10, no. 24, pp. 6531–6539,

2004.

[25] A. D. Villaverde, D. A. Donatti, and D. Bozinis, “Terbium gallium garnet

Verdet constant measurement with pulsed magnetic field,” Journal of Physics

C: Solid State Physics, vol. 11, pp. L495–498, 1978.

[26] P. A. Williams, A. H. Rose, G. W. Day, T. E. Milner, and M. N. Deeter,

“Temperature dependence of the Verdet constant in several diamagnetic

glasses,” Applied Optics, vol. 30, no. 10, pp. 1176–1178, 1991.

[27] N. P. Barnes and L. B. Petway, “Variation of the Verdet constant with tem-

perature of terbium gallium garnet,” Journal of Optical Society of America,

B, vol. 9, no. 10, pp. 1912–1915, 1992.

168

[28] E. Hecht, Optics, 4th ed. Addison Wesley Longman, Inc., 2002.

[29] W. Kaminsky, “Experimental and phenomenological aspects of circular bire-

fringence and related properties in transparent crystals,” Reports on Progress

in Physics, vol. 63, pp. 1575–1640, 2000.

[30] R. J. Potton, “Reciprocity in optics,” Reports on Progress in Physics, vol. 67,

pp. 717–754, 2004.

[31] G. Koeckelberghs, T. Foerier, T. Verbiest, and A. P. Persoons, “Chirality

and magnetic aspects of molecular nonlinear optics,” Proceedings of SPIE,

vol. 5935, 2005.

[32] O. Brevet-Philibert, R. Brunetton, and J. Monin, “Measuring the Verdet

constant: A simple, high precision, automatic device,” Journal of physics E:

Scientific Instruments, vol. 21, pp. 647–649, 1988.

[33] A. Jain, J. Kumar, F. Zhou, L. Li, and S. Tripathy, “A simple experiment

for determining Verdet constants using alternating current magnetic fields,”

American Journal of Physics, vol. 67, no. 8, pp. 714–717, 1999.

[34] C. C. Davis, “Building small, extremely sensitive practical interferometers

for sensor applications,” Nuclear Physics B, vol. 6, pp. 290–297, 1989.

[35] K. Cho, S. P. Bush, D. L. Mazzoni, and C. C. Davis, “Linear magnetic

birefringence measurement of Faraday materials,” Physical Review B, vol. 43,

no. 1, pp. 965–971, 1991.

169

[36] C. C. Davis, Lasers and Electro-Optics, Fundamentals and Engineering.

Cambridge: Cambridge University Press, 1996.

[37] A. Yariv, Optical Electronics in Modern Communications, 5th ed. New

York: Oxford University Press, 1996.

[38] M. Jhou, “Low-loss polymeric materials for passive waveguide components

in fiber optical telecommunication,” Optical Engineering, vol. 41, no. 7, pp.

1631–1643, 2002.

[39] L. Eldada and L. W. Shacklette, “Advances in polymer integrated optics,”

IEEE Journal of Selected Topics in Quantum Electronics, vol. 6, no. 1, pp.

54–68, 2000.

[40] S. M. Garner, V. Chuanov, S.-S. Lee, A. Chen, W. H. Steier, and L. R. Dal-

ton, “Vertically integrated waveguide polarization splitters using polymers,”

IEEE, Photonics Technology Letters, vol. 11, no. 7, pp. 842–844, 1999.

[41] Y. Wang, Y. Abe, Y. Matsuura, M. Miyagi, and H. Uyama, “Refractive

indices and extinction coefficients of polymers for the mid-infrared region,”

Applied Optics, vol. 37, no. 30, pp. 7091–7095, 1998.

[42] D. W. Smith Jr., S. Chen, S. M. Kumar, J. Ballato, C. Topping, H. V.

Shah, and S. Foulger, “Perfluorocyclobutyl copolymers for microphotonics,”

Advanced Materials, vol. 14, no. 21, pp. 1585–1589, 2002.

170

[43] J. Ballato, S. Foulger, and D. W. Smith Jr., “Optical properties of perflu-

orocyclobutyl polymers,” Journal of Optical Society of America, B, vol. 20,

no. 9, pp. 1838–1843, 2003.

[44] J. Ballato, S. H. Foulger, and D. W. Smith Jr., “Optical properties of perflu-

orocyclobutyl polymers. II. theoretical and experimental attenuation,” Jour-

nal of Optical Society of America, B, vol. 21, no. 5, pp. 958–967, 2004.

[45] B. M. A. Raman and J. B. Davies, “Vector-H finite element solution of

GaAs/GaAlAs rib waveguides,” IEE Proceedings J, vol. 132, no. 6, pp. 349–

353, 1985.

[46] Z.-E. Abid, K. L. Johnson, and A. Gopinath, “Analysis of dielectric guides

by vector transverse magnetic finite elements,” Journal of Lightwave Tech-

nology, vol. 11, no. 10, pp. 1545–1549, 1993.

[47] C. L. Xu, W. P. Huang, M. S. Stern, and S. K. Chaudhuri, “Full-vectorial

mode calculations by finite difference method,” IEE Proceedings J, vol. 141,

no. 5, pp. 281–286, 1994.

[48] S. Sujecki, T. M. Benson, P. Sewell, and P. C. Kendall, “Novel vectorial

analysis of optical waveguides,” Journal of Lightwave Technology, vol. 16,

no. 7, pp. 1329–1335, 1998.

171

[49] M. D. Feit and J. A. Flecker Jr, “Computation of mode properties in optical

fiber waveguides by a propagation beam method,” Applied Optics, vol. 19,

pp. 1154–1164, 1980.

[50] D. Yevick and B. Hermansson, “New formulations of the matrix beam prop-

agation method: application to rib waveguides,” IEEE Journal of Quantum

Electronics, vol. 25, no. 2, pp. 221–229, 1989.

[51] C. L. Xu, W. P. Huang, and S. K. Chaudhuri, “Efficient and accurate vec-

tor mode calculations by beam propagation method,” Journal of Lightwave

Technology, vol. 11, no. 7, pp. 1209–1215, 1993.

[52] T. E. Murphy, “Design, fabrication and measurement of integrated bragg

grating optical filters,” Ph.D, Massachusetts Institute of Technology, 2001.

[53] A. W. Snyder and J. D. Love, Optical Waveguide Theory, 1st ed. London:

Chapman and Hall, Ltd., 1983.

[54] R. R. Rye, A. J. Howard, and A. J. Ricco, “Photolithographic metallization

of fluorinated polymers,” Thin Solid Films, vol. 262, no. 1-2, pp. 73–83, 1995.

[55] K. Tanaka, T. Inomata, and M. Kogoma, “Improvement in adhesive strength

of fluorinated polymer films by atmospheric pressure glow plasma,” Thin

Solid Films, vol. 386, no. 2, pp. 217–221, 2001.

[56] Form No. 618-00200-498, Processing procedures for Dry-Etch CYCLOTENE

Advanced Electronics Resins (Dry-Etch BCB). Dow Chemical Co.

172

[57] R. G. Hunsperger, Integrate Optics: Theory and Technology, 4th ed. New

York: Springer, 1995.

[58] M. Haruna, Y. Segawa, and H. Nishihara, “Nondestructive and simple

method of optical waveguide loss measurement with optimisation of end-fire

coupling.” Electronics Letters, vol. 28, no. 17, pp. 1612–1613, 1992.

[59] R. G. Walker, “Simple and accurate loss measurement technique for semicon-

ductor optical waveguides,” Electronics Letters, vol. 21, no. 13, pp. 581–583,

1985.

[60] R. J. Deri and E. Kapon, “Low loss III-V semiconductor optical waveguides,”

IEEE Journal of Quantum Electronics, vol. 27, no. 3, pp. 626–640, 1991.

[61] N. Agarwal, S. Ponoth, J. Plawsky, and P. D. Persans, “Optimized oxygen

plasma etching of polyimide films for low loss optical waveguides,” Journal

of Vacuum Science and Technology A: Vacuum, Surfaces, and Films, vol. 20,

no. 5, pp. 1587–1591, 2002.

[62] T. Barwicz and H. A. Haus, “Three-dimensional analysis of scattering losses

due to sidewall roughness in microphotonic waveguides,” IEEE Journal of

Lightwave Technology, vol. 23, no. 9, pp. 2719–2732, 2005.

[63] T. Alder, A. Stohr, R. Heinzelmann, and D. Jager, “High-efficiency fiber-

to-chip coupling using low-loss tapered single-mode fiber,” IEEE Photonics

Technology Letters, vol. 12, no. 8, pp. 1016–1018, 2000.

173

[64] K. J. Vahala, “Optical microcavities,” Nature, vol. 424, no. 6950, pp. 839–

846, 2003.

[65] B. E. Little, S. T. Chu, H. A. Haus, J. Foresi, and J. P. Laine, “Microring

resonator channel dropping filters,” Journal of Lightwave Technology, vol. 15,

no. 6, pp. 998–1005, 1997.

[66] P. P. Absil, J. V. Hryniewicz, B. E. Little, R. A. Wilson, L. G. Joneckis, and

P.-T. Ho, “Compact microring notch filters,” IEEE Photonics Technology

Letters, vol. 12, no. 4, pp. 398–400, 2000.

[67] J. E. Heebner, “Nonlinear optical whispering gallery microresonators for pho-

tonics,” Ph.D. dissertation, University of Rochester, 2003.

[68] R. Grover, “Indium phosphide based optical micro-ring resonators.” Ph.D,

University of Maryland, College Park, 2003.

[69] H. A. Haus, Waves and Fields in Optoelectronics. Englewood Cliffs:

Prentice-Hall, Inc., 1984.

[70] D. Marcuse, “Bend loss of slab and fiber modes computed with diffraction

theory,” IEEE Journal of Quantum Electronics, vol. 29, no. 12, pp. 2957–

2961, 1993.

[71] T. E. Murphy, “Personal communications.”

174

[72] N.-. N. Feng, G.-. R. Zhou, C. Xu, and W. P. Huang, “Computation of

full-vector modes for bending waveguide using cylindrical perfectly matched

layers,” IEEE Journal of Lightwave Technology, vol. 20, no. 11, pp. 1976–

1980, 2002.

[73] W. P. Huang, C. L. Xu, and K. Yokoyama, “The perfectly matched layer

boundary condtion for modal analysis of optical waveguides: Leaky mode

calculations,” IEEE Photonics Technology Letters, vol. 8, no. 5, pp. 652–

654, 1996.

[74] H. A. Haus, W. P. Huang, S. Kawakami, and N. A. Whitaker, “Coupled-

mode theory of optical waveguides,” Journal of Lightwave Technology, vol.

LT-5, no. 1, pp. 16–23, 1987.

[75] W. Y. Chen, R. Grover, T. A. Ibrahim, V. Van, W. N. Herman, and P. T.

Ho, “High-finesse laterally coupled single-mode benzocyclobutene microring

resonators,” IEEE Photonics Technology Letters, vol. 16, no. 2, pp. 470–472,

2004.

[76] W. Cao, W. Y. Chen, J. Goldhar, P. T. Ho, W. N. Herman, and C. H. Lee,

“Optical bistability and picosecond optical switching in bisbenzocyclobutene

(BCB) polymer microring resonators,” in Optical Fiber Communication Con-

ference, 2004., 2004.

175

[77] C. H. Lee, P. S. Mak, and A. P. DeFonzo, “Optical control of millimeter-wave

propagation in dielectric waveguides,” IEEE Journal of Quantum Electron-

ics, vol. QE-16, no. 3, pp. 277–288, 1980.

[78] C. H. Lee, Picosecond Optoelectronic Devices. London: Academic Press,

1984.

[79] V. Van, T. A. Ibrahim, K. Ritter, P. P. Absil, F. G. Johnson, R. Grover,

J. Goldhar, and P. T. Ho, “All-optical nonlinear switching in GaAs-AlGaAs

microring resonators,” IEEE Photonics Technology Letters, vol. 14, no. 1,

pp. 74–76, 2002.

[80] V. Van, T. A. Ibrahim, P. P. Absil, F. G. Johnson, R. Grover, and P. T.

Ho, “Optical signal processing using nonlinear semiconductor microring res-

onators,” IEEE Journal of Selected Topics in Quantum Electronics, vol. 8,

no. 3, pp. 705–713, 2002.

[81] T. A. Ibrahim, W. Cao, Y. Kim, J. Li, J. Goldhar, P. T. Ho, and C. H. Lee,

“All-optical switching in a laterally coupled microring resonator by carrier

injection,” IEEE Photonics Technology Letters, vol. 15, no. 1, pp. 36–38,

2003.

[82] W. Cao, M. Du, Y. Kim, W. N. Herman, and C. H. Lee, “Ultrafast transient

photoconductivity of bisbenzocyclobutene (BCB) polymer films,” in OSA,

Ultrafast Electronics and Optoelectronics, 2003, pp. 49–52.

176

[83] T. A. Ibrahim, W. Cao, Y. Kim, J. Li, J. Goldhar, P. T. Ho, and C. H.

Lee, “Lightwave switching in semiconductor microring devices by free carrier

injection,” IEEE Journal of Lightwave Technology, vol. 21, no. 12, pp. 2997–

3003, 2003.

[84] V. Van, W. N. Herman, and H. Ping-Tong, “Linearized microring-loaded

Mach-Zehnder modulator with RF gain,” Journal of Lightwave Technology,

vol. 24, no. 4, pp. 1850–1854, 2006.

[85] J. H. Burroughes, D. D. C. Bradley, A. R. Brown, A. R. Marks, K. Mackay,

R. H. Friend, P. L. Burns, and A. B. Holmes, “Light-emitting diodes based

on conjugated polymers,” Nature, vol. 347, no. 6293, pp. 539–541, 1990.

[86] D. Braun and A. J. Heeger, “Visible light emission from semiconducting

polymer diodes,” Applied Physics Letters, vol. 58, no. 18, pp. 1982–1984,

1991.

[87] C. W. Tang, “Two-layer organic photovoltaic cell,” Applied Physics Letters,

vol. 48, no. 2, pp. 183–185, 1986.

[88] D. Wohrle and D. Meissner, “Organic solar cells,” Advanced Materials, vol. 3,

no. 3, pp. 129–138, 1991.

[89] C. J. Brabec, N. S. Sariciftci, and J. C. Hummelen, “Plastic solar cells,”

Advanced Functional Materials, vol. 11, no. 1, pp. 15–26, 2001.

177

[90] H. Hoppe and N. S. Sariciftci, “Organic solar cells: An overview,” Journal

of Material Research, vol. 19, no. 7, pp. 1924–1945, 2004.

[91] P. Peumans, A. Yakimov, and S. R. Forrest, “Small molecular weight organic

thin-film photodetectors and solar cells,” Journal of Applied Physics, vol. 93,

no. 7, pp. 3693–3723, 2003.

[92] P. Schilinsky, C. Waldauf, and C. J. Brabec, “Recombination and loss anal-

ysis in polythiophene based bulk heterojunction photodetectors,” Applied

Physics Letters, vol. 81, no. 20, pp. 3885–3887, 2002.

[93] G. Yu, Y. Cao, J. Wang, J. McElvain, and A. J. Heeger, “High sensitivity

polymer photosensors for image sensing applications,” Synthetic Metals, vol.

102, no. 1-3, pp. 904–907, 1999.

[94] P. Schilinsky, C. Waldauf, J. Hauch, and C. J. Brabec, “Polymer photovoltaic

detectors: progress and recent developments,” Thin Solid Films, vol. 451-452,

pp. 105–108, 2004.

[95] A. Tsumura, H. Koezuka, and T. Ando, “Macromolecular electronic device:

Field-effect transistor with a polythiophene thin film,” Applied Physics Let-

ters, vol. 49, no. 18, pp. 1210–1212, 1986.

[96] F. Ebisawa, T. Kurokawa, and S. Nara, “Electrical properties of polyacety-

lene/polysiloxane interface,” Journal of Applied Physics, vol. 54, no. 6, pp.

3255–3259, 1983.

178

[97] J. H. Burroughes, C. A. Jones, and R. H. Friend, “New semiconductor device

physics in polymer diodes and transitors,” Nature, vol. 335, no. 6186, pp.

137–141, 1988.

[98] T. W. Kelley, D. V. Muyres, P. F. Baude, T. P. Smith, and T. D. Jones,

“High performance organic thin film transistor,” in Material Research Society

Symposium Proceeding, vol. 771, 2003, pp. L6.5.1–L6.5.11.

[99] J. J. M. Halls, K. Pichler, R. H. Friend, S. C. Moratti, and A. B. Holmes,

“Exciton diffusion and dissociation in a poly(p-phenylenevinylene)/C60 het-

erojunction photovoltaic cell,” Applied Physics Letters, vol. 68, no. 22, pp.

3120–3122, 1996.

[100] T. Stubinger and W. Brutting, “Exciton diffusion and optical interference

in organic donor–acceptor photovoltaic cells,” Journal of Applied Physics,

vol. 90, no. 7, pp. 3632–3641, 2001.

[101] L. A. A. Pettersson, L. S. Roman, and O. Inganas, “Modeling photocurrent

action spectra of photovoltaic devices based on organic thin films,” Journal

of Applied Physics, vol. 86, no. 1, pp. 487–496, 1999.

[102] N. S. Sariciftci, L. Smilowitz, A. J. Heeger, and F. Wudl, “Photoinduced

eletron transfer from a conducting polymer to buckminsterfullerene,” Sci-

ence, vol. 258, no. 5087, pp. 1474–1476, 1992.

179

[103] C. H. Lee, G. Yu, and A. J. Heeger, “Persistent photoconductivity in poly(p-

phenylenevinylene): Spectral response and slow relaxation,” Physical Review

B, vol. 47, no. 23, pp. 15 543–15 553, 1993.

[104] L. Smilowitz, N. S. Sariciftci, R. Wu, C. Gettinger, A. J. Heeger, and F. Wudl,

“Photoexcitation spectroscopy of conducting-polymer/C60 composites: Pho-

toinduced electron transfer,” Physical Review B, vol. 47, no. 20, pp. 13 835–

13 842, 1993.

[105] K. Lee, R. A. J. Janssen, N. S. Sariciftci, and A. J. Heeger, “Direct evidence

of photoinduced electron transfer in conducting-polymer C60 composites by

infrared photoexcitation spectroscopy,” Physical Review B, vol. 49, no. 8, pp.

5781–5784, 1994.

[106] B. Kraabel, D. McBranch, N. S. Sariciftci, D. Moses, and A. J. Heeger, “Ul-

trafast spectroscopic studies of photoinduced electron transfer from semi-

conducting polymers to C60,” Physical Review B, vol. 50, no. 24, pp. 18 543–

18 552, 1994.

[107] M. C. J. M. Vissenberg and M. Matters, “Theory of the field-effect mobility

in amorphous organic transistors,” Physical Review B, vol. 57, no. 20, pp.

12 964–12 967, 1998.

[108] J. S. Kim, R. H. Friend, and F. Cacialli, “Improved operational stabil-

ity of polyfluorene-based organic light-emitting diodes with plasma-treated

180

indium–tin–oxide anodes,” Applied Physics Letters, vol. 74, no. 21, pp. 3084–

3086, 1999.

[109] T. M. Brown, J. S. Kim, R. H. Friend, F. Cacialli, R. Daik, and W. J. Feast,

“Built-in field electroabsorption spectroscopy of polymer light-emitting

diodes incorporating a doped poly(3,4-ethylene dioxythiophene) hole injec-

tion layer,” Applied Physics Letters, vol. 75, no. 12, pp. 1679–1681, 1999.

[110] P. K. H. Ho, J.-S. Kim, J. H. Burroughes, H. Becker, S. F. Y. Li, T. M.

Brown, F. Cacialli, and R. H. Friend, “Molecular-scale interface engineering

for polymer light-emitting diodes,” Nature, vol. 404, no. 6777, pp. 481–484,

2000.

[111] S. E. Shaheen, C. J. Brabec, N. S. Sariciftci, F. Padinger, T. Fromherz, and

J. C. Hummelen, “2.5% efficient organic plastic solar cells,” Applied Physics

Letters, vol. 78, no. 6, p. 841, 2001.

[112] R. P. Barber, M. Breban, R. D. Gomez, W. N. Herman, and D. B. Romero,

“Organic photocells based on block copolymer /C60 blends,” Proceedings of

SPIE, vol. 5938, pp. 341–348, 2005.

[113] R. P. Barber, R. D. Gomez, W. N. Herman, and D. B. Romero, “Characteri-

zation of organic photovoltaic devices based on a block copolymer / fullerene

blend,” Organic Electronics, to appear.

181

[114] L. S. Hung, C. W. Tang, and M. G. Mason, “Enhanced electron injection

in organic electroluminescence devices using an Al/LiF electrode,” Applied

Physics Letters, vol. 70, no. 2, p. 152, 1997.

[115] G. E. Jabbour, Y. Kawabe, S. E. Shaheen, J. F. Wang, M. M. Morrell,

B. Kippelen, and N. Peyghambarian, “Highly efficient and bright organic

electroluminescent devices with an aluminum cathode,” Applied Physics Let-

ters, vol. 71, no. 13, pp. 1762–1764, 1997.

[116] A. L. Farrenbruch and R. H. Bube, Fundamentals of Solar Cells: Photo-

voltaic Solar Energy Conversion. San Diego: Academic Press, 1983.

[117] A. D. Rakic, A. B. Djurisic, J. M. Elazar, and M. L. Majewski, “Optical

properties of metallic films for vertical-cavity optoelectronic devices,” Applied

Optics, vol. 37, no. 22, pp. 5271–5283, 1998.

[118] M. A. Ordal, L. L. Long, R. J. Bell, S. E. Bell, R. R. Bell, R. R. Alexander,

and C. A. Ward, “Optical properties of the metals Al, Co, Cu, Au, Fe, Pb,

Ni, Pd, Pt, Ag, Ti, and W in the infrared and far infrared,” Applied Optics,

vol. 22, no. 7, pp. 1099–1120, 1983.

[119] R. A. Synowicki, “Spectroscopic ellipsometry characterization of indium tin

oxide film microstructure and optical constants,” Thin Solid Films, vol. 313-

314, pp. 394–397, 1998.

182

[120] A. K. Ghosh and T. Feng, “Merocynanine organic solar cells,” Journal of

Applied Physics, vol. 49, no. 12, p. 5982, 1978.

[121] A. Desormeaux, J. J. Max, and R. M. Leblane, “Photovoltaic and electrical

properties of Al/Langmuir-Blodgett Films/Ag sandwich cells incorporating

either chlorophyll a, chlorophyll b, or zinc porphyrin derivative,” Journal of

Physical Chemistry, vol. 97, pp. 6670–6678, 1993.

[122] S. M. Sze, Physics of Semiconductor Devices. New York: Jone Wiley &

Sons, 1981, vol. 2nd.

[123] L. J. A. Koster, E. C. P. Smits, V. D. Mihailetchi, and P. W. M. Blom,

“Device model for the operation of polymer/fullerene bulk heterojunction

solar cells,” Physical Review B (Condensed Matter and Materials Physics),

vol. 72, no. 8, 2005.

183